Caveats concerning the ability of personality to predict job performance have been raised because of seemingly modest criterion-related validity. The goal of the present research was to test whether narrowing the context via the type of job (i.e., jobs with complex task demands) and adding a social skill-related moderator (i.e., political skill) would improve performance prediction. Further, along with political skill, the broad factor of personality demonstrated in prior research to have the strongest criterion validity (i.e., conscientiousness) was joined with a narrow construct closely related to openness to experience (i.e., learning approach) in a three-way interactive prediction of supervisor-rated task performance. With the employee–supervisor dyads among professionals, but not with the control group of non-professional employees, task performance was predicted by the three-way interaction, such that those high on all three received the highest performance ratings. Implications, strengths and limitations, and directions for future research are discussed. Copyright © 2012 John Wiley & Sons, Ltd.