We investigated whether the speed of responding to personality questionnaires affects the psychometric quality of the resulting questionnaire scores. Austrian adolescents (n = 106) completed Eysenck's Personality Inventory in a computerized version, which assessed single-item response latencies and the trait levels for neuroticism (N) and extraversion (E). We examined, first, the impact of questionnaire response instructions (speeded versus unspeeded) on test-retest reliability (7 weeks apart) and validity (self-peer agreement). A speeded instruction shortened response latencies, as expected, and made questionnaire scores more reliable (but not more valid). Second, we examined individual differences in habitual response speed. For N, fast responders showed greater validity (but not greater reliability) than slow responders; for E, no such effect was found. We explain these results by means of self-schemata, which appear to moderate the validity of personality assessments, but may do so differently for different traits.