In the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; American Psychiatric Association, 2013) personality disorder trait model, maladaptive behavior is located at one end of continuous scales. Widiger and colleagues, however, have argued that maladaptive behavior exists at both ends of trait continua. We propose that the role of evaluative variance differentiates these two perspectives and that once evaluation is isolated, maladaptive behaviors emerge at both ends of nonevaluative trait dimensions. In Study 1, we argue that evaluative variance is worthwhile to measure separately from descriptive content because it clusters items by valence regardless of content (e.g., lazy and workaholic; apathetic and anxious; gullible and paranoid; timid and hostile, etc.), which is unlikely to describe a consistent behavioral style. We isolate evaluation statistically (Study 2) and at the time of measurement (Study 3) to show that factors unrelated to valence evidence maladaptive behavior at both ends. We argue that nonevaluative factors, which display maladaptive behavior at both ends of continua, may better approximate ways in which individuals actually behave.