Unproctored, web-based assessments supposedly reduce social desirability distortions in self-report questionnaires because of an increased sense of privacy among participants. Three random-effects meta-analyses focusing either on social desirability (k = 30, total N = 3,746), the Big Five of personality (k = 66, total N = 2,951), or psychopathology (k = 96, total N = 16,034) compared social desirability distortions of self-reports across computerized and paper-and-pencil administration modes. Overall, a near-zero effect, = 0.01, was obtained that did not indicate less socially desirable responding in computerized assessments. Moreover, moderator analyses did not identify differential effects for proctored and unproctored procedures. Thus, paper-and-pencil and computerized administrations of self-report scales yield comparable mean scores. Unproctored web-based surveys do not offer an advantage with regard to socially desirable responding in self-report questionnaires.