Looman and Marshall (2005) suggested that phallometric assessment data have low reliability. It is correct that reliability places a ceiling on validity. This creases an interesting but false paradox. The authors argue that the only possible answer to this paradox is that traditional measures of reliability underestimate the true reliability of phallometric data; for example, the day-to-day variability of overall arousability likely affects the reliability of indexes that do not compensate for it. The relative arousal to rape scenarios is the variable that most consistently distinguishes rapists from other men, including offenders who have not committed sex offenses; and as such is one of the most important constructs for theory development and clinical assessment regarding rape. The so-called strong form of the sexual preference hypothesis states that rapists prefer rape over consenting sex and thus should show more sexual arousal to rape than consenting stories in phallometric assessment. Looman and Marshall apparently used all deviant stimuli to calculate the differential rape indexes for the Quinsey et al. (1981) stimulus set, including, presumably, the nonsexual violence stories. Because rapists as a group typically show a small response to this stimulus category and because the strong form of the sexual preference hypothesis says nothing about responses to nonsexual violence, this is not a fair test of the hypothesis. 21 references