diff --git a/AUTHORS.rst b/AUTHORS.rst
index ae3d941db4..33339da13c 100644
--- a/AUTHORS.rst
+++ b/AUTHORS.rst
@@ -19,6 +19,7 @@ their individual contributions.
 * `Alex Stapleton <https://www.github.com/public>`_
 * `Alex Willmer <https://github.com/moreati>`_ (alex@moreati.org.uk)
 * `Andrea Pierré <https://www.github.com/kir0ul>`_
+* `Andrea Reina <https://www.github.com/andreareina>`_
 * `Ben Anhalt <https://github.com/benanhalt>`_
 * `Ben Peterson <https://github.com/killthrush>`_ (killthrush@hotmail.com)
 * `Benjamin Lee <https://github.com/Benjamin-Lee>`_ (benjamindlee@me.com)
diff --git a/hypothesis-python/RELEASE.rst b/hypothesis-python/RELEASE.rst
new file mode 100644
index 0000000000..78dc7b0da1
--- /dev/null
+++ b/hypothesis-python/RELEASE.rst
@@ -0,0 +1,5 @@
+RELEASE_TYPE: patch
+
+This patch clarifies the reporting of time spent generating data. A
+simple arithmetic mean of the percentage of time spent can be
+misleading; reporting the actual time spent avoids misunderstandings.
diff --git a/hypothesis-python/src/hypothesis/statistics.py b/hypothesis-python/src/hypothesis/statistics.py
index 73377077cd..bdf56466e4 100644
--- a/hypothesis-python/src/hypothesis/statistics.py
+++ b/hypothesis-python/src/hypothesis/statistics.py
@@ -9,7 +9,6 @@
 # obtain one at https://mozilla.org/MPL/2.0/.
 
 import math
-import statistics
 from collections import Counter
 
 from hypothesis.utils.dynamicvariables import DynamicVariable
@@ -41,6 +40,24 @@ def describe_targets(best_targets):
         return lines
 
 
+def format_ms(times):
+    """Format `times` into a string representing approximate milliseconds.
+
+    `times` is a collection of durations in seconds.
+    """
+    ordered = sorted(times)
+    n = max(0, len(ordered) - 1)
+    lower = int(ordered[int(math.floor(n * 0.05))] * 1000)
+    upper = int(ordered[int(math.ceil(n * 0.95))] * 1000)
+    if upper == 0:
+        ms = "< 1ms"
+    elif lower == upper:
+        ms = f"~ {lower}ms"
+    else:
+        ms = f"~ {lower}-{upper} ms"
+    return ms
+
+
 def describe_statistics(stats_dict):
     """Return a multi-line string describing the passed run statistics.
 
@@ -63,22 +80,11 @@ def describe_statistics(stats_dict):
         if not cases:
             continue
         statuses = Counter(t["status"] for t in cases)
-        runtimes = sorted(t["runtime"] for t in cases)
-        n = max(0, len(runtimes) - 1)
-        lower = int(runtimes[int(math.floor(n * 0.05))] * 1000)
-        upper = int(runtimes[int(math.ceil(n * 0.95))] * 1000)
-        if upper == 0:
-            ms = "< 1ms"
-        elif lower == upper:
-            ms = f"~ {lower}ms"
-        else:
-            ms = f"{lower}-{upper} ms"
-        drawtime_percent = 100 * statistics.mean(
-            t["drawtime"] / t["runtime"] if t["runtime"] > 0 else 0 for t in cases
-        )
+        runtime_ms = format_ms(t["runtime"] for t in cases)
+        drawtime_ms = format_ms(t["drawtime"] for t in cases)
         lines.append(
             f"  - during {phase} phase ({d['duration-seconds']:.2f} seconds):\n"
-            f"    - Typical runtimes: {ms}, ~ {drawtime_percent:.0f}% in data generation\n"
+            f"    - Typical runtimes: {runtime_ms}, of which {drawtime_ms} in data generation\n"
             f"    - {statuses['valid']} passing examples, {statuses['interesting']} "
             f"failing examples, {statuses['invalid'] + statuses['overrun']} invalid examples"
         )
diff --git a/hypothesis-python/tests/cover/test_statistical_events.py b/hypothesis-python/tests/cover/test_statistical_events.py
index 56b3733769..8e889aeaeb 100644
--- a/hypothesis-python/tests/cover/test_statistical_events.py
+++ b/hypothesis-python/tests/cover/test_statistical_events.py
@@ -8,6 +8,7 @@
 # v. 2.0. If a copy of the MPL was not distributed with this file, You can
 # obtain one at https://mozilla.org/MPL/2.0/.
 
+import re
 import time
 import traceback
 
@@ -145,7 +146,7 @@ def test(i):
 
 @pytest.mark.parametrize("draw_delay", [False, True])
 @pytest.mark.parametrize("test_delay", [False, True])
-def test_draw_time_percentage(draw_delay, test_delay):
+def test_draw_timing(draw_delay, test_delay):
     time.freeze()
 
     @st.composite
@@ -161,11 +162,10 @@ def test(_):
 
     stats = describe_statistics(call_for_statistics(test))
     if not draw_delay:
-        assert "~ 0%" in stats
-    elif test_delay:
-        assert "~ 50%" in stats
+        assert "< 1ms" in stats
     else:
-        assert "~ 100%" in stats
+        match = re.search(r"of which ~ (?P<gentime>\d+)", stats)
+        assert 49 <= int(match.group("gentime")) <= 51
 
 
 def test_has_lambdas_in_output():