diff --git a/docs/user_guide.md b/docs/user_guide.md
index 9cbaeae..9fd4ac1 100644
--- a/docs/user_guide.md
+++ b/docs/user_guide.md
@@ -140,8 +140,8 @@
 (or set `BENCHMARK_OUT`). Specify the output format with
 `--benchmark_out_format={json|console|csv}` (or set
 `BENCHMARK_OUT_FORMAT={json|console|csv}`). Note that the 'csv' reporter is
-deprecated and the saved `.csv` file 
-[is not parsable](https://github.com/google/benchmark/issues/794) by csv 
+deprecated and the saved `.csv` file
+[is not parsable](https://github.com/google/benchmark/issues/794) by csv
 parsers.
 
 Specifying `--benchmark_out` does not suppress the console output.
@@ -983,6 +983,25 @@
   ->Arg(512);
 ```
 
+While usually the statistics produce values in time units,
+you can also produce percentages:
+
+```c++
+void BM_spin_empty(benchmark::State& state) {
+  for (auto _ : state) {
+    for (int x = 0; x < state.range(0); ++x) {
+      benchmark::DoNotOptimize(x);
+    }
+  }
+}
+
+BENCHMARK(BM_spin_empty)
+  ->ComputeStatistics("ratio", [](const std::vector<double>& v) -> double {
+    return std::begin(v) / std::end(v);
+  }, benchmark::StatisticUnit::Percentage)
+  ->Arg(512);
+```
+
 <a name="using-register-benchmark" />
 
 ## Using RegisterBenchmark(name, fn, args...)
diff --git a/include/benchmark/benchmark.h b/include/benchmark/benchmark.h
index 31f2b02..fbb5340 100644
--- a/include/benchmark/benchmark.h
+++ b/include/benchmark/benchmark.h
@@ -450,6 +450,8 @@
 
 typedef uint64_t IterationCount;
 
+enum StatisticUnit { kTime, kPercentage };
+
 // BigOFunc is passed to a benchmark in order to specify the asymptotic
 // computational complexity for the benchmark.
 typedef double(BigOFunc)(IterationCount);
@@ -462,9 +464,11 @@
 struct Statistics {
   std::string name_;
   StatisticsFunc* compute_;
+  StatisticUnit unit_;
 
-  Statistics(const std::string& name, StatisticsFunc* compute)
-      : name_(name), compute_(compute) {}
+  Statistics(const std::string& name, StatisticsFunc* compute,
+             StatisticUnit unit = kTime)
+      : name_(name), compute_(compute), unit_(unit) {}
 };
 
 class BenchmarkInstance;
@@ -965,7 +969,8 @@
   Benchmark* Complexity(BigOFunc* complexity);
 
   // Add this statistics to be computed over all the values of benchmark run
-  Benchmark* ComputeStatistics(std::string name, StatisticsFunc* statistics);
+  Benchmark* ComputeStatistics(std::string name, StatisticsFunc* statistics,
+                               StatisticUnit unit = kTime);
 
   // Support for running multiple copies of the same benchmark concurrently
   // in multiple threads.  This may be useful when measuring the scaling
@@ -1421,6 +1426,7 @@
 
     Run()
         : run_type(RT_Iteration),
+          aggregate_unit(kTime),
           error_occurred(false),
           iterations(1),
           threads(1),
@@ -1444,6 +1450,7 @@
     int64_t per_family_instance_index;
     RunType run_type;
     std::string aggregate_name;
+    StatisticUnit aggregate_unit;
     std::string report_label;  // Empty if not set by benchmark.
     bool error_occurred;
     std::string error_message;
diff --git a/src/benchmark_register.cc b/src/benchmark_register.cc
index 0a90be3..56fef5c 100644
--- a/src/benchmark_register.cc
+++ b/src/benchmark_register.cc
@@ -399,8 +399,9 @@
 }
 
 Benchmark* Benchmark::ComputeStatistics(std::string name,
-                                        StatisticsFunc* statistics) {
-  statistics_.emplace_back(name, statistics);
+                                        StatisticsFunc* statistics,
+                                        StatisticUnit unit) {
+  statistics_.emplace_back(name, statistics, unit);
   return this;
 }
 
diff --git a/src/complexity.cc b/src/complexity.cc
index 7757315..b87697f 100644
--- a/src/complexity.cc
+++ b/src/complexity.cc
@@ -199,6 +199,7 @@
   big_o.repetition_index = Run::no_repetition_index;
   big_o.threads = reports[0].threads;
   big_o.aggregate_name = "BigO";
+  big_o.aggregate_unit = StatisticUnit::kTime;
   big_o.report_label = reports[0].report_label;
   big_o.iterations = 0;
   big_o.real_accumulated_time = result_real.coef;
@@ -220,6 +221,7 @@
   rms.per_family_instance_index = reports[0].per_family_instance_index;
   rms.run_type = BenchmarkReporter::Run::RT_Aggregate;
   rms.aggregate_name = "RMS";
+  rms.aggregate_unit = StatisticUnit::kPercentage;
   rms.report_label = big_o.report_label;
   rms.iterations = 0;
   rms.repetition_index = Run::no_repetition_index;
diff --git a/src/console_reporter.cc b/src/console_reporter.cc
index 6fd7645..de2a7a0 100644
--- a/src/console_reporter.cc
+++ b/src/console_reporter.cc
@@ -142,10 +142,16 @@
   } else if (result.report_rms) {
     printer(Out, COLOR_YELLOW, "%10.0f %-4s %10.0f %-4s ", real_time * 100, "%",
             cpu_time * 100, "%");
-  } else {
+  } else if (result.run_type != Run::RT_Aggregate ||
+             result.aggregate_unit == StatisticUnit::kTime) {
     const char* timeLabel = GetTimeUnitString(result.time_unit);
     printer(Out, COLOR_YELLOW, "%s %-4s %s %-4s ", real_time_str.c_str(), timeLabel,
             cpu_time_str.c_str(), timeLabel);
+  } else {
+    assert(result.aggregate_unit == StatisticUnit::kPercentage);
+    printer(Out, COLOR_YELLOW, "%10.2f %-4s %10.2f %-4s ",
+            (100. * result.real_accumulated_time), "%",
+            (100. * result.cpu_accumulated_time), "%");
   }
 
   if (!result.report_big_o && !result.report_rms) {
diff --git a/src/json_reporter.cc b/src/json_reporter.cc
index f3ddfb9..22d5ce0 100644
--- a/src/json_reporter.cc
+++ b/src/json_reporter.cc
@@ -251,6 +251,15 @@
   out << indent << FormatKV("threads", run.threads) << ",\n";
   if (run.run_type == BenchmarkReporter::Run::RT_Aggregate) {
     out << indent << FormatKV("aggregate_name", run.aggregate_name) << ",\n";
+    out << indent << FormatKV("aggregate_unit", [&run]() -> const char* {
+      switch (run.aggregate_unit) {
+        case StatisticUnit::kTime:
+          return "time";
+        case StatisticUnit::kPercentage:
+          return "percentage";
+      }
+      BENCHMARK_UNREACHABLE();
+    }()) << ",\n";
   }
   if (run.error_occurred) {
     out << indent << FormatKV("error_occurred", run.error_occurred) << ",\n";
@@ -258,8 +267,17 @@
   }
   if (!run.report_big_o && !run.report_rms) {
     out << indent << FormatKV("iterations", run.iterations) << ",\n";
-    out << indent << FormatKV("real_time", run.GetAdjustedRealTime()) << ",\n";
-    out << indent << FormatKV("cpu_time", run.GetAdjustedCPUTime());
+    if (run.run_type != Run::RT_Aggregate ||
+        run.aggregate_unit == StatisticUnit::kTime) {
+      out << indent << FormatKV("real_time", run.GetAdjustedRealTime())
+          << ",\n";
+      out << indent << FormatKV("cpu_time", run.GetAdjustedCPUTime());
+    } else {
+      assert(run.aggregate_unit == StatisticUnit::kPercentage);
+      out << indent << FormatKV("real_time", run.real_accumulated_time)
+          << ",\n";
+      out << indent << FormatKV("cpu_time", run.cpu_accumulated_time);
+    }
     out << ",\n"
         << indent << FormatKV("time_unit", GetTimeUnitString(run.time_unit));
   } else if (run.report_big_o) {
diff --git a/src/statistics.cc b/src/statistics.cc
index 5f82c27..88dfc35 100644
--- a/src/statistics.cc
+++ b/src/statistics.cc
@@ -155,6 +155,7 @@
     data.repetitions = reports[0].repetitions;
     data.repetition_index = Run::no_repetition_index;
     data.aggregate_name = Stat.name_;
+    data.aggregate_unit = Stat.unit_;
     data.report_label = report_label;
 
     // It is incorrect to say that an aggregate is computed over
@@ -167,13 +168,15 @@
     data.real_accumulated_time = Stat.compute_(real_accumulated_time_stat);
     data.cpu_accumulated_time = Stat.compute_(cpu_accumulated_time_stat);
 
-    // We will divide these times by data.iterations when reporting, but the
-    // data.iterations is not nessesairly the scale of these measurements,
-    // because in each repetition, these timers are sum over all the iterations.
-    // And if we want to say that the stats are over N repetitions and not
-    // M iterations, we need to multiply these by (N/M).
-    data.real_accumulated_time *= iteration_rescale_factor;
-    data.cpu_accumulated_time *= iteration_rescale_factor;
+    if (data.aggregate_unit == StatisticUnit::kTime) {
+      // We will divide these times by data.iterations when reporting, but the
+      // data.iterations is not necessarily the scale of these measurements,
+      // because in each repetition, these timers are sum over all the iters.
+      // And if we want to say that the stats are over N repetitions and not
+      // M iterations, we need to multiply these by (N/M).
+      data.real_accumulated_time *= iteration_rescale_factor;
+      data.cpu_accumulated_time *= iteration_rescale_factor;
+    }
 
     data.time_unit = reports[0].time_unit;
 
diff --git a/test/complexity_test.cc b/test/complexity_test.cc
index 0de73c5..7407359 100644
--- a/test/complexity_test.cc
+++ b/test/complexity_test.cc
@@ -36,6 +36,7 @@
        {"\"repetitions\": %int,$", MR_Next},
        {"\"threads\": 1,$", MR_Next},
        {"\"aggregate_name\": \"BigO\",$", MR_Next},
+       {"\"aggregate_unit\": \"time\",$", MR_Next},
        {"\"cpu_coefficient\": %float,$", MR_Next},
        {"\"real_coefficient\": %float,$", MR_Next},
        {"\"big_o\": \"%bigo\",$", MR_Next},
@@ -49,6 +50,7 @@
        {"\"repetitions\": %int,$", MR_Next},
        {"\"threads\": 1,$", MR_Next},
        {"\"aggregate_name\": \"RMS\",$", MR_Next},
+       {"\"aggregate_unit\": \"percentage\",$", MR_Next},
        {"\"rms\": %float$", MR_Next},
        {"}", MR_Next}});
   AddCases(TC_CSVOut, {{"^\"%bigo_name\",,%float,%float,%bigo,,,,,$"},
diff --git a/test/repetitions_test.cc b/test/repetitions_test.cc
index 2f94d05..569777d 100644
--- a/test/repetitions_test.cc
+++ b/test/repetitions_test.cc
@@ -59,6 +59,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"mean\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -73,6 +74,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"median\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -87,6 +89,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"stddev\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -164,6 +167,7 @@
                        {"\"repetitions\": 3,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"mean\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": %int,$", MR_Next},
                        {"\"real_time\": %float,$", MR_Next},
                        {"\"cpu_time\": %float,$", MR_Next},
@@ -177,6 +181,7 @@
                        {"\"repetitions\": 3,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"median\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": %int,$", MR_Next},
                        {"\"real_time\": %float,$", MR_Next},
                        {"\"cpu_time\": %float,$", MR_Next},
@@ -190,6 +195,7 @@
                        {"\"repetitions\": 3,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"stddev\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": %int,$", MR_Next},
                        {"\"real_time\": %float,$", MR_Next},
                        {"\"cpu_time\": %float,$", MR_Next},
diff --git a/test/reporter_output_test.cc b/test/reporter_output_test.cc
index 989eb48..7c7d994 100644
--- a/test/reporter_output_test.cc
+++ b/test/reporter_output_test.cc
@@ -1,5 +1,6 @@
 
 #undef NDEBUG
+#include <numeric>
 #include <utility>
 
 #include "benchmark/benchmark.h"
@@ -454,6 +455,7 @@
                        {"\"repetitions\": 2,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"mean\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 2,$", MR_Next},
                        {"\"name\": \"BM_Repeat/repeats:2_median\",$"},
                        {"\"family_index\": 15,$", MR_Next},
@@ -463,6 +465,7 @@
                        {"\"repetitions\": 2,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"median\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 2,$", MR_Next},
                        {"\"name\": \"BM_Repeat/repeats:2_stddev\",$"},
                        {"\"family_index\": 15,$", MR_Next},
@@ -472,6 +475,7 @@
                        {"\"repetitions\": 2,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"stddev\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 2,$", MR_Next}});
 ADD_CASES(TC_CSVOut, {{"^\"BM_Repeat/repeats:2\",%csv_report$"},
                       {"^\"BM_Repeat/repeats:2\",%csv_report$"},
@@ -519,6 +523,7 @@
                        {"\"repetitions\": 3,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"mean\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 3,$", MR_Next},
                        {"\"name\": \"BM_Repeat/repeats:3_median\",$"},
                        {"\"family_index\": 16,$", MR_Next},
@@ -528,6 +533,7 @@
                        {"\"repetitions\": 3,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"median\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 3,$", MR_Next},
                        {"\"name\": \"BM_Repeat/repeats:3_stddev\",$"},
                        {"\"family_index\": 16,$", MR_Next},
@@ -537,6 +543,7 @@
                        {"\"repetitions\": 3,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"stddev\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 3,$", MR_Next}});
 ADD_CASES(TC_CSVOut, {{"^\"BM_Repeat/repeats:3\",%csv_report$"},
                       {"^\"BM_Repeat/repeats:3\",%csv_report$"},
@@ -594,6 +601,7 @@
                        {"\"repetitions\": 4,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"mean\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 4,$", MR_Next},
                        {"\"name\": \"BM_Repeat/repeats:4_median\",$"},
                        {"\"family_index\": 17,$", MR_Next},
@@ -603,6 +611,7 @@
                        {"\"repetitions\": 4,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"median\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 4,$", MR_Next},
                        {"\"name\": \"BM_Repeat/repeats:4_stddev\",$"},
                        {"\"family_index\": 17,$", MR_Next},
@@ -612,6 +621,7 @@
                        {"\"repetitions\": 4,$", MR_Next},
                        {"\"threads\": 1,$", MR_Next},
                        {"\"aggregate_name\": \"stddev\",$", MR_Next},
+                       {"\"aggregate_unit\": \"time\",$", MR_Next},
                        {"\"iterations\": 4,$", MR_Next}});
 ADD_CASES(TC_CSVOut, {{"^\"BM_Repeat/repeats:4\",%csv_report$"},
                       {"^\"BM_Repeat/repeats:4\",%csv_report$"},
@@ -661,6 +671,7 @@
            {"\"repetitions\": 3,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"mean\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 3,$", MR_Next},
            {"\"name\": \"BM_SummaryRepeat/repeats:3_median\",$"},
            {"\"family_index\": 19,$", MR_Next},
@@ -670,6 +681,7 @@
            {"\"repetitions\": 3,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"median\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 3,$", MR_Next},
            {"\"name\": \"BM_SummaryRepeat/repeats:3_stddev\",$"},
            {"\"family_index\": 19,$", MR_Next},
@@ -679,6 +691,7 @@
            {"\"repetitions\": 3,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"stddev\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 3,$", MR_Next}});
 ADD_CASES(TC_CSVOut, {{".*BM_SummaryRepeat/repeats:3 ", MR_Not},
                       {"^\"BM_SummaryRepeat/repeats:3_mean\",%csv_report$"},
@@ -709,6 +722,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"mean\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 2,$", MR_Next},
            {"\"name\": \"BM_SummaryDisplay/repeats:2_median\",$"},
            {"\"family_index\": 20,$", MR_Next},
@@ -718,6 +732,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"median\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 2,$", MR_Next},
            {"\"name\": \"BM_SummaryDisplay/repeats:2_stddev\",$"},
            {"\"family_index\": 20,$", MR_Next},
@@ -727,6 +742,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"stddev\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 2,$", MR_Next}});
 ADD_CASES(TC_CSVOut,
           {{".*BM_SummaryDisplay/repeats:2 ", MR_Not},
@@ -761,6 +777,7 @@
            {"\"repetitions\": 3,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"mean\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 3,$", MR_Next},
            {"\"time_unit\": \"us\",?$"},
            {"\"name\": \"BM_RepeatTimeUnit/repeats:3_median\",$"},
@@ -771,6 +788,7 @@
            {"\"repetitions\": 3,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"median\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 3,$", MR_Next},
            {"\"time_unit\": \"us\",?$"},
            {"\"name\": \"BM_RepeatTimeUnit/repeats:3_stddev\",$"},
@@ -781,6 +799,7 @@
            {"\"repetitions\": 3,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"stddev\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 3,$", MR_Next},
            {"\"time_unit\": \"us\",?$"}});
 ADD_CASES(TC_CSVOut,
@@ -869,6 +888,7 @@
      {"\"repetitions\": 3,$", MR_Next},
      {"\"threads\": 1,$", MR_Next},
      {"\"aggregate_name\": \"mean\",$", MR_Next},
+     {"\"aggregate_unit\": \"time\",$", MR_Next},
      {"\"iterations\": 3,$", MR_Next},
      {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next},
      {"\"name\": \"BM_UserStats/iterations:5/repeats:3/manual_time_median\",$"},
@@ -880,6 +900,7 @@
      {"\"repetitions\": 3,$", MR_Next},
      {"\"threads\": 1,$", MR_Next},
      {"\"aggregate_name\": \"median\",$", MR_Next},
+     {"\"aggregate_unit\": \"time\",$", MR_Next},
      {"\"iterations\": 3,$", MR_Next},
      {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next},
      {"\"name\": \"BM_UserStats/iterations:5/repeats:3/manual_time_stddev\",$"},
@@ -891,6 +912,7 @@
      {"\"repetitions\": 3,$", MR_Next},
      {"\"threads\": 1,$", MR_Next},
      {"\"aggregate_name\": \"stddev\",$", MR_Next},
+     {"\"aggregate_unit\": \"time\",$", MR_Next},
      {"\"iterations\": 3,$", MR_Next},
      {"\"real_time\": %float,$", MR_Next},
      {"\"name\": \"BM_UserStats/iterations:5/repeats:3/manual_time_\",$"},
@@ -902,6 +924,7 @@
      {"\"repetitions\": 3,$", MR_Next},
      {"\"threads\": 1,$", MR_Next},
      {"\"aggregate_name\": \"\",$", MR_Next},
+     {"\"aggregate_unit\": \"time\",$", MR_Next},
      {"\"iterations\": 3,$", MR_Next},
      {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next}});
 ADD_CASES(
@@ -917,6 +940,154 @@
      {"^\"BM_UserStats/iterations:5/repeats:3/manual_time_\",%csv_report$"}});
 
 // ========================================================================= //
+// ------------- Testing relative standard deviation statistics ------------ //
+// ========================================================================= //
+
+const auto UserPercentStatistics = [](const std::vector<double>& v) {
+  return 1. / 100.;
+};
+void BM_UserPercentStats(benchmark::State& state) {
+  for (auto _ : state) {
+    state.SetIterationTime(150 / 10e8);
+  }
+}
+// clang-format off
+BENCHMARK(BM_UserPercentStats)
+  ->Repetitions(3)
+  ->Iterations(5)
+  ->UseManualTime()
+  ->Unit(benchmark::TimeUnit::kNanosecond)
+  ->ComputeStatistics("", UserPercentStatistics, benchmark::StatisticUnit::kPercentage);
+// clang-format on
+
+// check that UserPercent-provided stats is calculated, and is after the
+// default-ones empty string as name is intentional, it would sort before
+// anything else
+ADD_CASES(TC_ConsoleOut,
+          {{"^BM_UserPercentStats/iterations:5/repeats:3/manual_time [ "
+            "]* 150 ns %time [ ]*5$"},
+           {"^BM_UserPercentStats/iterations:5/repeats:3/manual_time [ "
+            "]* 150 ns %time [ ]*5$"},
+           {"^BM_UserPercentStats/iterations:5/repeats:3/manual_time [ "
+            "]* 150 ns %time [ ]*5$"},
+           {"^BM_UserPercentStats/iterations:5/repeats:3/"
+            "manual_time_mean [ ]* 150 ns %time [ ]*3$"},
+           {"^BM_UserPercentStats/iterations:5/repeats:3/"
+            "manual_time_median [ ]* 150 ns %time [ ]*3$"},
+           {"^BM_UserPercentStats/iterations:5/repeats:3/"
+            "manual_time_stddev [ ]* 0.000 ns %time [ ]*3$"},
+           {"^BM_UserPercentStats/iterations:5/repeats:3/manual_time_ "
+            "[ ]* 1.00 % [ ]* 1.00 %[ ]*3$"}});
+ADD_CASES(
+    TC_JSONOut,
+    {{"\"name\": \"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$"},
+     {"\"family_index\": 23,$", MR_Next},
+     {"\"per_family_instance_index\": 0,$", MR_Next},
+     {"\"run_name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$",
+      MR_Next},
+     {"\"run_type\": \"iteration\",$", MR_Next},
+     {"\"repetitions\": 3,$", MR_Next},
+     {"\"repetition_index\": 0,$", MR_Next},
+     {"\"threads\": 1,$", MR_Next},
+     {"\"iterations\": 5,$", MR_Next},
+     {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next},
+     {"\"name\": \"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$"},
+     {"\"family_index\": 23,$", MR_Next},
+     {"\"per_family_instance_index\": 0,$", MR_Next},
+     {"\"run_name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$",
+      MR_Next},
+     {"\"run_type\": \"iteration\",$", MR_Next},
+     {"\"repetitions\": 3,$", MR_Next},
+     {"\"repetition_index\": 1,$", MR_Next},
+     {"\"threads\": 1,$", MR_Next},
+     {"\"iterations\": 5,$", MR_Next},
+     {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next},
+     {"\"name\": \"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$"},
+     {"\"family_index\": 23,$", MR_Next},
+     {"\"per_family_instance_index\": 0,$", MR_Next},
+     {"\"run_name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$",
+      MR_Next},
+     {"\"run_type\": \"iteration\",$", MR_Next},
+     {"\"repetitions\": 3,$", MR_Next},
+     {"\"repetition_index\": 2,$", MR_Next},
+     {"\"threads\": 1,$", MR_Next},
+     {"\"iterations\": 5,$", MR_Next},
+     {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next},
+     {"\"name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time_mean\",$"},
+     {"\"family_index\": 23,$", MR_Next},
+     {"\"per_family_instance_index\": 0,$", MR_Next},
+     {"\"run_name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$",
+      MR_Next},
+     {"\"run_type\": \"aggregate\",$", MR_Next},
+     {"\"repetitions\": 3,$", MR_Next},
+     {"\"threads\": 1,$", MR_Next},
+     {"\"aggregate_name\": \"mean\",$", MR_Next},
+     {"\"aggregate_unit\": \"time\",$", MR_Next},
+     {"\"iterations\": 3,$", MR_Next},
+     {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next},
+     {"\"name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time_median\",$"},
+     {"\"family_index\": 23,$", MR_Next},
+     {"\"per_family_instance_index\": 0,$", MR_Next},
+     {"\"run_name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$",
+      MR_Next},
+     {"\"run_type\": \"aggregate\",$", MR_Next},
+     {"\"repetitions\": 3,$", MR_Next},
+     {"\"threads\": 1,$", MR_Next},
+     {"\"aggregate_name\": \"median\",$", MR_Next},
+     {"\"aggregate_unit\": \"time\",$", MR_Next},
+     {"\"iterations\": 3,$", MR_Next},
+     {"\"real_time\": 1\\.5(0)*e\\+(0)*2,$", MR_Next},
+     {"\"name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time_stddev\",$"},
+     {"\"family_index\": 23,$", MR_Next},
+     {"\"per_family_instance_index\": 0,$", MR_Next},
+     {"\"run_name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$",
+      MR_Next},
+     {"\"run_type\": \"aggregate\",$", MR_Next},
+     {"\"repetitions\": 3,$", MR_Next},
+     {"\"threads\": 1,$", MR_Next},
+     {"\"aggregate_name\": \"stddev\",$", MR_Next},
+     {"\"aggregate_unit\": \"time\",$", MR_Next},
+     {"\"iterations\": 3,$", MR_Next},
+     {"\"real_time\": %float,$", MR_Next},
+     {"\"name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time_\",$"},
+     {"\"family_index\": 23,$", MR_Next},
+     {"\"per_family_instance_index\": 0,$", MR_Next},
+     {"\"run_name\": "
+      "\"BM_UserPercentStats/iterations:5/repeats:3/manual_time\",$",
+      MR_Next},
+     {"\"run_type\": \"aggregate\",$", MR_Next},
+     {"\"repetitions\": 3,$", MR_Next},
+     {"\"threads\": 1,$", MR_Next},
+     {"\"aggregate_name\": \"\",$", MR_Next},
+     {"\"aggregate_unit\": \"percentage\",$", MR_Next},
+     {"\"iterations\": 3,$", MR_Next},
+     {"\"real_time\": 1\\.(0)*e-(0)*2,$", MR_Next}});
+ADD_CASES(TC_CSVOut, {{"^\"BM_UserPercentStats/iterations:5/repeats:3/"
+                       "manual_time\",%csv_report$"},
+                      {"^\"BM_UserPercentStats/iterations:5/repeats:3/"
+                       "manual_time\",%csv_report$"},
+                      {"^\"BM_UserPercentStats/iterations:5/repeats:3/"
+                       "manual_time\",%csv_report$"},
+                      {"^\"BM_UserPercentStats/iterations:5/repeats:3/"
+                       "manual_time_mean\",%csv_report$"},
+                      {"^\"BM_UserPercentStats/iterations:5/repeats:3/"
+                       "manual_time_median\",%csv_report$"},
+                      {"^\"BM_UserPercentStats/iterations:5/repeats:3/"
+                       "manual_time_stddev\",%csv_report$"},
+                      {"^\"BM_UserPercentStats/iterations:5/repeats:3/"
+                       "manual_time_\",%csv_report$"}});
+
+// ========================================================================= //
 // ------------------------- Testing StrEscape JSON ------------------------ //
 // ========================================================================= //
 #if 0  // enable when csv testing code correctly handles multi-line fields
diff --git a/test/user_counters_tabular_test.cc b/test/user_counters_tabular_test.cc
index 421f27b..ff81e26 100644
--- a/test/user_counters_tabular_test.cc
+++ b/test/user_counters_tabular_test.cc
@@ -125,6 +125,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"mean\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -146,6 +147,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"median\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -167,6 +169,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"stddev\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -231,6 +234,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 2,$", MR_Next},
            {"\"aggregate_name\": \"median\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -252,6 +256,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 2,$", MR_Next},
            {"\"aggregate_name\": \"stddev\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": %int,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
diff --git a/test/user_counters_thousands_test.cc b/test/user_counters_thousands_test.cc
index bbe1942..a42683b 100644
--- a/test/user_counters_thousands_test.cc
+++ b/test/user_counters_thousands_test.cc
@@ -96,6 +96,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"mean\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 2,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -115,6 +116,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"median\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 2,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
@@ -134,6 +136,7 @@
            {"\"repetitions\": 2,$", MR_Next},
            {"\"threads\": 1,$", MR_Next},
            {"\"aggregate_name\": \"stddev\",$", MR_Next},
+           {"\"aggregate_unit\": \"time\",$", MR_Next},
            {"\"iterations\": 2,$", MR_Next},
            {"\"real_time\": %float,$", MR_Next},
            {"\"cpu_time\": %float,$", MR_Next},
diff --git a/tools/gbench/Inputs/test4_run0.json b/tools/gbench/Inputs/test4_run0.json
new file mode 100644
index 0000000..54cf127
--- /dev/null
+++ b/tools/gbench/Inputs/test4_run0.json
@@ -0,0 +1,21 @@
+{
+  "context": {
+    "date": "2016-08-02 17:44:46",
+    "num_cpus": 4,
+    "mhz_per_cpu": 4228,
+    "cpu_scaling_enabled": false,
+    "library_build_type": "release"
+  },
+  "benchmarks": [
+    {
+      "name": "whocares",
+      "run_type": "aggregate",
+      "aggregate_name": "zz",
+      "aggregate_unit": "percentage",
+      "iterations": 1000,
+      "real_time": 0.01,
+      "cpu_time": 0.10,
+      "time_unit": "ns"
+    }
+  ]
+}
diff --git a/tools/gbench/Inputs/test4_run1.json b/tools/gbench/Inputs/test4_run1.json
new file mode 100644
index 0000000..25d5605
--- /dev/null
+++ b/tools/gbench/Inputs/test4_run1.json
@@ -0,0 +1,21 @@
+{
+  "context": {
+    "date": "2016-08-02 17:44:46",
+    "num_cpus": 4,
+    "mhz_per_cpu": 4228,
+    "cpu_scaling_enabled": false,
+    "library_build_type": "release"
+  },
+  "benchmarks": [
+    {
+      "name": "whocares",
+      "run_type": "aggregate",
+      "aggregate_name": "zz",
+      "aggregate_unit": "percentage",
+      "iterations": 1000,
+      "real_time": 0.005,
+      "cpu_time": 0.15,
+      "time_unit": "ns"
+    }
+  ]
+}
diff --git a/tools/gbench/report.py b/tools/gbench/report.py
index 504e15f..8203cba 100644
--- a/tools/gbench/report.py
+++ b/tools/gbench/report.py
@@ -914,6 +914,69 @@
             assert_measurements(self, out, expected)
 
 
+
+class TestReportDifferenceForPercentageAggregates(
+        unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        def load_results():
+            import json
+            testInputs = os.path.join(
+                os.path.dirname(
+                    os.path.realpath(__file__)),
+                'Inputs')
+            testOutput1 = os.path.join(testInputs, 'test4_run0.json')
+            testOutput2 = os.path.join(testInputs, 'test4_run1.json')
+            with open(testOutput1, 'r') as f:
+                json1 = json.load(f)
+            with open(testOutput2, 'r') as f:
+                json2 = json.load(f)
+            return json1, json2
+
+        json1, json2 = load_results()
+        cls.json_diff_report = get_difference_report(
+            json1, json2, utest=True)
+
+    def test_json_diff_report_pretty_printing(self):
+        expect_lines = [
+            ['whocares', '-0.5000', '+0.5000', '0', '0', '0', '0']
+        ]
+        output_lines_with_header = print_difference_report(
+            self.json_diff_report,
+            utest=True, utest_alpha=0.05, use_color=False)
+        output_lines = output_lines_with_header[2:]
+        print("\n")
+        print("\n".join(output_lines_with_header))
+        self.assertEqual(len(output_lines), len(expect_lines))
+        for i in range(0, len(output_lines)):
+            parts = [x for x in output_lines[i].split(' ') if x]
+            self.assertEqual(expect_lines[i], parts)
+
+    def test_json_diff_report(self):
+        expected_output = [
+            {
+                'name': u'whocares',
+                'measurements': [
+                    {'time': -0.5,
+                     'cpu': 0.5,
+                     'real_time': 0.01,
+                     'real_time_other': 0.005,
+                     'cpu_time': 0.10,
+                     'cpu_time_other': 0.15}
+                ],
+                'time_unit': 'ns',
+                'utest': {}
+            }
+        ]
+        self.assertEqual(len(self.json_diff_report), len(expected_output))
+        for out, expected in zip(
+                self.json_diff_report, expected_output):
+            self.assertEqual(out['name'], expected['name'])
+            self.assertEqual(out['time_unit'], expected['time_unit'])
+            assert_utest(self, out, expected)
+            assert_measurements(self, out, expected)
+
+
 class TestReportSorting(unittest.TestCase):
     @classmethod
     def setUpClass(cls):