Social Desirability (SD) scales are sometimes treated, by researchers, as measures of dishonesty and, by practitioners, as indicators of faking on self-report assessments in high-stakes settings, such as personnel selection. Applying SD scales to measure dishonesty or faking, however, remains a point of contention among the scientific community. This two-part study investigated if SD scales, with a True/False response format, are valid for these purposes. Initially, 46 participants completed an SD scale and 12 personality items while under instruction to “think aloud”, that is, to verbalize all the thoughts they had. These spoken thoughts were recorded and transcribed. Next, 175 judges rated the participants' honesty in relation to each SD item, based on the participants' transcribed spoken thoughts and their selected response to the item. The results showed that responses keyed as “socially desirable responding” were judged as significantly less honest than those not keyed as such. However, the effect size was very small, and the socially desirable responses were still being judged as somewhat honest overall. Further, participants' SD scale sum scores were not related to the judges' ratings of participant honesty on the personality items. Thus, overall, SD scales appear to be a poor measure of dishonesty.