There is widespread agreement that response bias variables, such as good impression, have little relevance, either as moderators or suppressors, in efforts to improve the accuracy of employment‐related predictions based on personality test scores. However, a recent review found that there were relatively few methodologically sound real‐life studies of the moderator function. Reported here are four such studies involving the prediction of job performance. As hypothesized, a significant moderator effect was shown in the two studies that utilized ‘transparent’ job‐related predictors, with useful prediction only at the lower levels of good impression, but not in the two studies that utilized personality variables as predictors. These results confirm that validity when using transparent items to predict employment‐related success can be affected by the operation of good‐impression response bias. Further research is needed to delineate other relevant effects of good impression in such situations.