A second-order meta-analysis was conducted to assess the implications of using college student subjects in social science research. Four meta-analyses investigating response homogeneity (cumulative N > 650,000) and 30 meta-analyses reporting effect sizes for 65 behavioral or psychological relationships (cumulative N > 350,000) provided comparative data for college student subjects and nonstudent (adult) subjects for the present research. In general, responses of college student subjects were found to be slightly more homogeneous than those of nonstudent subjects. Moreover, effect sizes derived from college student subjects frequently differed from those derived from nonstudent subjects both directionally and in magnitude. Because there was no systematic pattern to the differences observed, caution must be exercised when attempting to extend any relationship found using college student subjects to a nonstudent (adult) population. The results augur in favor of, and emphasize the importance of, replicating research based on college student subjects with nonstudent subjects before attempting any generalizations. Copyright 2001 by the University of Chicago.