diff --git a/gstlal-inspiral/python/stats/inspiral_lr.py b/gstlal-inspiral/python/stats/inspiral_lr.py
index 58f9ceabd7c0acaae06125aa7abed160e4a7b342..13e4e77844cd94ba9d3e5eb22933eb569200f7a9 100644
--- a/gstlal-inspiral/python/stats/inspiral_lr.py
+++ b/gstlal-inspiral/python/stats/inspiral_lr.py
@@ -729,8 +729,15 @@ class LnNoiseDensity(LnLRDensity):
 		lnP += self.coinc_rates.lnP_instruments(**triggers_per_second_per_template)[frozenset(snrs)]
 
 		# evaluate dt and dphi parameters
-		# NOTE: uniform and normalized so that the log should be zero, but there is no point in doing that
-		# lnP += 0
+		# NOTE: this assumes uniform in dt and dphi.  The dt part is
+		# approximate and should be detector dependent and involve a
+		# slightly intracate calculation since coincidence must be
+		# mutual, but we igore that in this normalization and get it
+		# close enough.
+		# ( 1  / \Deta T ) ( 1 / \Delta \phi) for each detector
+		# hardcode \Delta T = 20 ms
+		#
+		lnP += (1. / 0.020 * 1. / (2. * numpy.pi))**(len(snrs) - 1)
 
 		# evaluate the rest
 		interps = self.interps