In recent years, several techniques have been developed to measure implicit social cognition. Despite their increased use, little attention has been devoted to their reliability and validity. This paper reports a direct assessment of the inter-item consistency, stability, and convergent validity of some implicit attitude measures. Black-White race attitudes were measured on four separate occasions, each two weeks apart, using three implicit measures (Response-Window Evaluative Priming, Implicit Association Test, and Response-Window IAT) and one explicit measure (Modern Racism Scale). After correcting for inter-item inconsistency with latent variable analyses, we see that (a) stability indices are improved and (b) implicit measures are substantially correlated with each other, forming a single latent factor. The psychometric properties of response-latency implicit measures are stronger than recently suggested.