McFarland and Crouch (2002) reported substantial positive correlations (a) between the Implicit Association Test (IAT) and response speed and (b) between IATs assessing racism or self-esteem and ostensibly unrelated control IATs. Using an IAT measure in millisecond-difference score format, they concluded that the IAT was confounded with general cognitive ability. A reanalysis of these data using the D measure (Greenwald, Nosek, & Banaji, 2003) eliminated the speed of responding confound, although it did not eliminate the correlation between the control and racism IATs. The study was replicated and the two correlations, paralleling those in the original study, emerged for the millisecond-difference score. However, both were reduced to nonsignificance by use of the D measure. These findings are consistent with other recent studies (Mierke & Klauer, 2003) that document the protection afforded by D against cognitive skill confounds.