Experiment 1 used the Implicit Association Test (IAT; A. G. Greenwald, D. E
. McGhee, Be J. L. K. Schwartz, 1998) to measure self-esteem by assessing a
utomatic associations of self with positive or negative valence. Confirmato
ry factor analysis (CFA) showed that two IAT measures defined a factor that
was distinct from, but weakly correlated with, a factor defined by standar
d explicit (self-report) measures of self-esteem. Experiment 2 tested known
-groups validity of two IAT gender self-concept measures. Compared with wel
l-established explicit measures, the IAT measures revealed triple the diffe
rence in measured masculinity-femininity between men and women. Again, CFA
revealed construct divergence between implicit and explicit measures. Exper
iment 3 assessed the self-esteem IAT's validity in predicting cognitive rea
ctions to success and failure. High implicit self-esteem was associated in
the predicted fashion with buffering against adverse effects of failure on
two of four measures.