OBJECTIVE: to investigate the association between variability and loss of body weight with subsequent cognitive performance and activities of daily living in older individuals. DESIGN: cross-sectional cohort study. SETTING: PROspective Study of Pravastatin in the Elderly at Risk, multicentre trial with participants from Scotland, Ireland and the Netherlands. SUBJECTS: 4,309 participants without severe cognitive dysfunction (mean age 75.1 years, standard deviation (SD) = 3.3), at higher risk for cardiovascular disease (CVD). METHODS: body weight was measured every 3 months for 2.5 years. Weight loss was defined as an average slope across all weight measurements and as ≥ 5% decrease in baseline body weight during follow-up. Visit-to-visit variability was defined as the SD of weight measurements (kg) between visits. Four tests of cognitive function were examined: Stroop test, letter-digit coding test (LDCT), immediate and delayed picture-word learning tests. Two measures of daily living activities: Barthel Index (BI) and instrumental activities of daily living (IADL). All tests were examined at month 30. RESULTS: both larger body weight variability and loss of ≥ 5% of baseline weight were independently associated with worse scores on all cognitive tests, but minimally with BI and IADL. Compared with participants with stable weight, participants with significant weight loss performed 5.83 seconds (95% CI 3.74; 7.92) slower on the Stroop test, coded 1.72 digits less (95% CI -2.21; -1.13) on the LDCT and remembered 0.71 pictures less (95% CI -0.93; -0.48) on the delayed picture-word learning test. CONCLUSION: in older people at higher risk for CVD, weight loss and variability are independent risk-factors for worse cognitive function.