The study examined the Implicit Relational Assessment Procedure's (IRAP) validity as a computerized response-latency-based measure of implicit self-esteem. University undergraduates and 2 sets of convicted prisoners participated. One set of prisoners resided in the main block, and the other in a privileged lower security "open area" of a medium-security Irish prison. The IRAP required participants to maintain relational responses that were self-positive on half of the IRAP trials ("Consistent"), and self-negative on the other half ("Inconsistent"). As predicted, the students and the prisoners in the open area showed stronger IRAP effects (shorter latencies during consistent vs. inconsistent trials) than the main block prisoners. Additionally, the IRAP's convergent validity was supported by its moderate positive correlation with an explicit self-esteem measure. The findings provide preliminary support for the analytic utility of the IRAP and suggest future avenues of investigation afforded by the IRAP's design.