Erectile assessments of sexual preference have been found quite reliable. Their validity, however, has been questioned with some findings that are at variance with the clinical data. These apparently distorted results could be either false positive or false negative responses. The former may be due to faking, which is the act of feigning a spurious positive response to a non-preferred stimulus, and the latter could be from suppressing, which is producing a spurious negative response to one's natural sexual preference. In this study, phallometric testing with a mercury-in-rubber strain gauge and Model 240 Parks Plethysmograph was used in conjunction with a mid-lumbar pressure sensor. Twenty subjects, half of whom were sex offenders, were tested with visual and audiovisual stimuli depicting male and female models at various ages and degrees of sexual development. Other testing included self-report questionnaires and estimates of age and sexual development. It was found that both groups significantly reduced penile tumescence when instructed to suppress to their stated sexual preference in the audiovisual series. A significant reduction was also obtained in suppressing to the category to which they had a maximum response in the same series. When both groups were instructed to fake to a non-preferred stimulus, the levels of arousal generated were quite variable and not significantly different from those obtained before instruction. Extraneous movement as measured by the lumbar pressure device was markedly greater in our three incest offenders. The meaning of this finding is unclear. Sex offenders as a group were found to be significantly more variable in their responses than controls.(ABSTRACT TRUNCATED AT 250 WORDS)