Like person judgments in everyday life, diagnoses of personality disorders (PDs) reflect at least two influences: first, the actual characteristics of the target person (substance), and second, the more positive or negative view that the perceiver has of the target person (evaluation). In this article, we present a systematic account of substance and evaluation in PD diagnoses, using a modified version of Brunswik's (1956) lens model. An empirical study shows that the items of the alternative DSM-5 model for personality disorders (AMPD) are in fact saturated with evaluation, that this evaluation is largely the same as “social desirability,” and that the two mandatory components of the AMPD (criterion A and B) are not easily distinguishable from one another in that regard. We provide concrete recommendations as to how the conceptual clarity of PD diagnostics may be improved, by distinguishing people's personality dispositions from their (likely) long-term consequences.