diff --git a/src/lib/cycle/__tests__/prediction.test.ts b/src/lib/cycle/__tests__/prediction.test.ts
index 4e97d3963..08b994f4a 100644
--- a/src/lib/cycle/__tests__/prediction.test.ts
+++ b/src/lib/cycle/__tests__/prediction.test.ts
@@ -1,6 +1,7 @@
 import { describe, expect, it } from "vitest";
 
 import { addDays } from "../day-math";
+import { FERTILE_POST, FERTILE_PRE } from "../types";
 import {
   clampLuteal,
   confirmSymptothermal,
@@ -1091,3 +1092,33 @@ describe("cycle/dto — toCyclePredictionDTO still-learning honesty gate (M-1)",
     expect(goalGated.predictedOvulation).toBeNull();
   });
 });
+
+describe("cycle/prediction — the fertile window's width (A10)", () => {
+  it("spans seven days, ending the day after the ovulation estimate", () => {
+    // The comment on the constants used to say six days and the constants made
+    // seven. The width is now a decision with a reason written at the constant,
+    // so it is pinned here rather than left to be re-read off the same numbers
+    // the code uses: five days before the estimate, the estimate, and one after.
+    const cycles = cyclesFromGaps("2026-01-01", [28, 28, 28]);
+    const p = predictCycle(cycles, [], BASE_PROFILE, "2026-03-20", []);
+
+    expect(p.predictedOvulation).not.toBeNull();
+    expect(p.fertileWindowStart).toBe(addDays(p.predictedOvulation!, -5));
+    expect(p.fertileWindowEnd).toBe(addDays(p.predictedOvulation!, 1));
+
+    const span =
+      (Date.parse(`${p.fertileWindowEnd}T12:00:00Z`) -
+        Date.parse(`${p.fertileWindowStart}T12:00:00Z`)) /
+      86_400_000;
+    // Inclusive of both ends.
+    expect(span + 1).toBe(7);
+  });
+
+  it("keeps the extra day on the side that costs least to be wrong about", () => {
+    // The trailing day is estimator slack plus ovum viability. Dropping it
+    // would shorten the window for somebody avoiding pregnancy, which is the
+    // direction that misses.
+    expect(FERTILE_PRE).toBe(5);
+    expect(FERTILE_POST).toBe(1);
+  });
+});
diff --git a/src/lib/cycle/types.ts b/src/lib/cycle/types.ts
index 16eada87f..33983bc14 100644
--- a/src/lib/cycle/types.ts
+++ b/src/lib/cycle/types.ts
@@ -103,7 +103,26 @@ export const LUTEAL_DEFAULT = 14;
 export const LUTEAL_MIN = 10;
 export const LUTEAL_MAX = 16;
 
-/** §4 — six-day fertile window: ovulation − 5 … ovulation + 1. */
+/**
+ * §4 — the fertile window: ovulation − 5 … ovulation + 1, seven days.
+ *
+ * Wilcox's six-day window is the five days BEFORE ovulation plus the day
+ * itself, and the trailing day here is a deliberate departure from it. The
+ * six-day figure is measured against a KNOWN ovulation day; ours is estimated,
+ * usually by back-calculating a luteal length from a median cycle, and an
+ * estimate that lands a day early would put the last fertile day outside a
+ * window drawn tight to it. The ovum also stays viable for the better part of a
+ * day after release, so the extra day is not only estimator slack.
+ *
+ * The asymmetry of the cost decides the direction. This window is surfaced to
+ * people trying to conceive and to people avoiding it, and for the second group
+ * a window one day too short is a window that misses. The constants are the
+ * conservative reading; the comment used to claim six days and describe seven,
+ * which is the only part of this that was wrong.
+ *
+ * Changing either constant changes what the iOS client reimplements from the
+ * same spec, so it needs a coordination note, not just an edit here.
+ */
 export const FERTILE_PRE = 5;
 export const FERTILE_POST = 1;