A correlational study examined the validity of brief role-play tests for assessing social skills. 39 male and 39 female undergraduates responded to 20 role-played scenarios involving heterosocial interactions. They were surreptitiously observed 1 wk later while they interacted with an opposite-sex student who, in actuality, was an experimental confederate. Videotapes of the role-play test and a naturalistic interaction were subsequently rated for a variety of component responses. Correlations of responses in the 2 situations provided equivocal results. Role-play behavior was moderately correlated with behavior in the naturalistic situation for females, but there were few significant relationships for males. Results are discussed in terms of the validity and utility of role-play tests, and several unresolved issues are identified. (19 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1979 American Psychological Association.