ImportanceHigher prepandemic physical activity (PA) levels have been associated with lower risk and severity of COVID-19.ObjectiveTo investigate the association between self-reported prepandemic PA levels and the risk and severity of COVID-19 in older US adults.Design, Setting, and ParticipantsThis cohort study combined cohorts from 3 ongoing prospective randomized clinical trials of US adults aged 45 years or older who provided prepandemic self-reports of baseline leisure-time PA and risk factors for COVID-19 outcomes using the most recent questionnaire completed as of December 31, 2019, as the baseline PA assessment. In multiple surveys from May 2020 through May 2022, participants indicated whether they had at least 1 positive COVID-19 test result or were diagnosed with or hospitalized for COVID-19.ExposurePrepandemic PA, categorized into 3 groups by metabolic equivalent hours per week: inactive (0-3.5), insufficiently active (>3.5 to <7.5), and sufficiently active (≥7.5).Main Outcome and MeasuresPrimary outcomes were risk of COVID-19 and hospitalization for COVID-19. Multivariable logistic regression was used to estimate odd ratios (ORs) and 95% CIs for the association of COVID-19 diagnosis and/or hospitalization with each of the 2 upper PA categories vs the lowest PA category.ResultsThe pooled cohort included 61 557 participants (mean [SD] age, 75.7 [6.4] years; 70.7% female), 20.2% of whom were inactive; 11.4%, insufficiently active; and 68.5%, sufficiently active. A total of 5890 confirmed incident cases of COVID-19 were reported through May 2022, including 626 hospitalizations. After controlling for demographics, body mass index, lifestyle factors, comorbidities, and medications used, compared with inactive individuals, those insufficiently active had no significant reduction in infection (OR, 0.96; 95% CI, 0.86-1.06) or hospitalization (OR, 0.98; 95% CI, 0.76-1.28), whereas those sufficiently active had a significant reduction in infection (OR, 0.90; 95% CI, 0.84-0.97) and hospitalization (OR, 0.73; 95% CI, 0.60-0.90). In subgroup analyses, the association between PA and SARS-CoV-2 infection differed by sex, with only sufficiently active women having decreased odds (OR, 0.87; 95% CI, 0.79-0.95; P = .04 for interaction).Conclusions and RelevanceIn this cohort study of adults aged 45 years or older, those who adhered to PA guidelines before the pandemic had lower odds of developing or being hospitalized for COVID-19. Thus, higher prepandemic PA levels may be associated with reduced odds of SARS-CoV-2 infection and hospitalization for COVID-19.