diff --git a/petbox/dca/primary.py b/petbox/dca/primary.py
index ab701bc..0225264 100644
--- a/petbox/dca/primary.py
+++ b/petbox/dca/primary.py
@@ -21,7 +21,7 @@
 
 import numpy as np
 
-from scipy.special import expi as ei, gammainc  # type: ignore
+from scipy.special import expi as ei, gammainc, gamma  # type: ignore
 
 from abc import ABC, abstractmethod
 from typing import (TypeVar, Type, List, Dict, Tuple, Any,
@@ -983,7 +983,15 @@ def _Nfn(self, t: NDFloat, **kwargs: Any) -> NDFloat:
         qi = self.qi
         tau = self.tau
         n = self.n
-        return qi * tau / n * gammainc(1.0 / n, (t / tau) ** n)
+        # N(t) = qi tau / n * gamma(1/n) * P(1/n, (t/tau)^n), where P is the regularised
+        # lower incomplete gamma (scipy gammainc). The gamma(1/n) factor is required: without
+        # it the cumulative and EUR are wrong by a factor of gamma(1/n) (e.g. +33% at n=0.4).
+        coef = qi * tau / n * gamma(1.0 / n)
+        if np.isfinite(coef):
+            return coef * gammainc(1.0 / n, (t / tau) ** n)
+        # gamma(1/n) overflows for very small n (where the closed-form EUR diverges);
+        # fall back to the bounded numerical integral.
+        return self._integrate_with(self._qfn, t, **kwargs)
 
     def _Dfn(self, t: NDFloat) -> NDFloat:
         tau = self.tau
diff --git a/test/test_perf.py b/test/test_perf.py
index 91a78a6..a835074 100644
--- a/test/test_perf.py
+++ b/test/test_perf.py
@@ -12,6 +12,17 @@ def _quad_cum(rate_fn, t: np.ndarray) -> np.ndarray:
     return np.array([quad(rate_fn, 0.0, float(ti))[0] for ti in t])
 
 
+@pytest.mark.parametrize('n', [0.3, 0.4, 0.5, 0.6, 0.8])
+def test_SE_cum_matches_integral(n: float) -> None:
+    """SE.cum must equal the integral of SE.rate. Regression for the missing
+    gamma(1/n) factor, which made cum/EUR wrong by a factor of gamma(1/n)."""
+    qi, tau = 1000.0, 30.0
+    se = dca.SE(qi, tau, n)
+    t = dca.get_time(1.0, 5000.0, 60)
+    ref = _quad_cum(lambda s: qi * np.exp(-(s / tau) ** n), t)
+    assert np.allclose(se.cum(t), ref, rtol=1e-4)
+
+
 def test_integrate_with_PLE_accuracy() -> None:
     """Verify _integrate_with produces correct cumulative volumes for PLE."""
     ple = dca.PLE(qi=1000.0, Di=0.01, Dinf=0.0001, n=0.5)