Faking in self-report personality scales (SRPSs) is not sufficiently understood. This limits its detection and prevention. Here, we introduce a taxonomy of faking behaviors that constitute faking strategies in SRPSs, reflecting the stages (comprehension, retrieval, judgment, and response) of the general response process model (GRPM). We reanalyzed data from two studies investigating the faking of high and low scores on Extraversion (E) and Need for Cognition (NFC) scales (Data Set 1; N = 305) or on an E scale (Data Set 2; N = 251). Participants were asked to explain exactly what they did to fake, and their responses (N = 533) were examined via a qualitative content analysis. The resulting taxonomy included 22 global and 13 specific behaviors that (in combination) constitute faking strategies in SRPSs. We organized the behaviors into four clusters along the stages of the GRPM. The behaviors held irrespective of the construct (E or NFC), and with two exceptions, also irrespective of the data set (Data Sets 1 or 2). Eight exceptions concerning faking direction (high or low) indicate direction-specific differences in faking behaviors. Respondents reported using not only different faking behaviors (e.g., role-playing, behaviors to avoid being detected) but also multiple combinations thereof. The suggested taxonomy is necessarily limited to the specified context, and, thus, additional faking behaviors are possible. To fully understand faking, further research in other contexts should be conducted to complement the taxonomy. Still, the complexity shown here explains why adequate detection and prevention of faking in SRPSs is so challenging.