Introduction: Reduced physical functioning is a risk factor for many adverse outcomes for older adults, including risk of institutionalization and disability. Previous research indicates that leisure-time physical activity (LTPA) in mid-life predicts better physical function at a single timepoint in late-life, but more information is needed on the association between mid-life LTPA and trajectories of physical function in late-life. We hypothesized that higher levels of LTPA in mid-life are associated with attenuated physical function declines in late life. Methods: We studied 5,517 ARIC Study participants (Visit 5; mean age: 75.9 years; range: 67-91 years; 58% women; 21% Black participants) without Parkinson disease or prevalent stroke who completed the Short Physical Performance Battery (SPPB, a composite of gait speed, balance, and chair stands (lowest score 0, highest score 12). SPPB assessments were repeated at Visits 6 and 7 (median follow-up: 4.2 years). Engagement in LTPA was assessed using a modified Baecke Physical Activity Questionnaire in mid-life (Visit 1; mean age: 52 years; mean time between mid- and late-life exams: 23.7 years). Based on minutes/week of moderate-vigorous intensity LTPA participation, participants were classified into poor (0 min/week), intermediate (1-149), or ideal (≥150) LTPA levels, according to American Heart Association (AHA) physical activity recommendations. We also used the average total MET-mins/week and average moderate-vigorous intensity MET-min/week as continuous measures. Random-slope and random-intercept mixed models with an independent covariance structure tested the association between mid-life LTPA and SPPB score change in late-life, adjusted for age, sex, race-center, education, body mass index, smoking, diabetes, and hypertension. Results: SPPB scores declined an average of 2.19 points per 10 years (95% CI: -2.43, -1.96; p<0.001) among older adults who had poor mid-life LTPA. In contrast, participants with intermediate or ideal mid-life LTPA levels declined 0.45 SPPB points less (95% CI: 0.11, 0.78; p<0.001) and 0.65 SPPB points less (95% CI: 0.35, 0.95; p<0.001), respectively, per 10 years compared to those with poor mid-life LTPA. In continuous analyses, each additional standard deviation higher total MET-min/week PA was associated with a 0.24 slower SPPB point decline per 10 years (0.13, 0.36; p<0.001). Continuous results remained consistent when considering only moderate-vigorous intensity MET-min/week. Conclusions: Higher levels of self-reported LTPA at a single time point in mid-life were associated with mitigated physical function declines in SPPB in late-life, over two decades later. Additional research may consider the role of changes in LTPA across mid- and late-life and of other types of physical activity, as well as sedentary behaviors, in further understanding the association.