Recent evidence suggests that the Psychopathic Personality Inventory (PPI) is composed of two orthogonal factors, one representing traits such as dominance and low anxiety (PPI-I) and a second that reflects more socially deviant aspects of psychopathy (PPI-II). To assess whether response sets differentially affect these factors, the present research reanalyzed data from two dissimulation studies in which college students completed the PPI both honestly and with instructions either to "fake good" (n = 186) or "fake bad" (n = 143). PPI-II scores decreased significantly in the fake-good condition (d = .86) and increased significantly when faking bad (d = 2.65). PPI-I scores were unaffected by attempts to feign serious psychopathology (d = .10) and actually increased in the fake-good condition (d = .53). These findings significantly qualify previous conclusions about the effect of response distortion on the PPI.