The results of two studies are reported examining the utility of a pictorial-modified Stroop task (P-MST) in the assessment of sexual interest in a sample of nonoffending participants and of sexual offenders against children. A mixed factorial design was adopted for both. Nine gay and 12 straight participants took part in the first study which found that participants typically had attentional bias on the P-MST that was in line with their stated sexual interests. Twenty four sexual offenders against children and 24 control participants took part in the second study. Again results indicated that the task was tapping into the participants’ stated sexual interests. Furthermore, extrafamilial offenders and offenders with an admitted sexual interest in children demonstrated the greatest mean bias for child stimuli relative to adult stimuli. A cautious interpretation of the results was recommended, given the sample size in the study, the heterogeneity of the sample, differences in cognitive speed among offenders and controls and other methodological caveats.