To assess driving self-restriction (vision related and nonvision related) in relation to vision test performance of older adults. Cross-sectional study. Population-based cohort of community-dwelling older adults. Six hundred twenty-nine current drivers aged 55 and older had driving behavior, health, and physical function assessed and vision function tested in 1993-95. Self-report of driving restriction as vision or non-vision related and performance on a comprehensive battery of vision tests (visual acuity; contrast sensitivity; effects of illumination level, contrast, and glare on acuity; visual fields with and without attentional load; color vision; temporal sensitivity; and the effect of dim light on walking ability). Demographic, health, and functional characteristics differed significantly between restrictors and nonrestrictors but not between vision- and nonvision-related restrictors. Controlling for potential confounding, only vision-related driving self-restriction was significantly associated with reduced performance on nonstandard measures of acuity. Poor depth perception was significantly associated with restriction for both vision- and nonvision-related reasons. Poor performance on attentional visual field tests, analyzed individually and in combination with standard field tests, was not associated with driving self-restriction. Older adults with early changes in spatial vision function and depth perception appear to recognize their limitations and restrict their driving even if they do not acknowledge the visual impairment as the cause for restriction. Poor visual attention, a risk factor for crashes, may not be recognized. Additional studies of driving self-restriction in relation to risk factors for crashes in older adults may help refine this strategy of reducing driving-related injury and death.