Objective To evaluate the association of the Fat‐but‐Fit paradigm with cognitive function in an older adult population. Methods Data from the 1999–2002 National Health and Nutrition Examination Survey were used (N = 2,336 adults 60–85 yrs). Physical activity was assessed via open‐ended questions asking about participation in 48 leisure‐time activities over the previous 30 days. Using Metabolic Equivalent of Task calculations for each activity, participants were classified as either active or inactive. Participants were then classified into one of six groups (Normal BMI and Inactive [referent], Overweight BMI and Inactive, Obese BMI and Inactive, Normal BMI and Active, Overweight BMI and Active, Obese BMI and Active), based on their activity status and body mass index (BMI). The Digit Symbol Substitution Test (DSST) was employed to assess cognitive function. Results Compared to those who had a normal BMI and were inactive (referent), those who had a normal BMI and were active had a 4‐unit higher DSST score (β = 4.0; 95% CI: 1.0–6.9; P = 0.009); those with an overweight BMI and active had a 4.5‐unit higher DSST score (β = 4.5; 95% CI: 1.3–7.6; P = 0.008); and those who had an obese BMI and active had a 3.7‐unit higher DSST score (β = 3.7; 95% CI: 0.96–6.4; P = 0.01). Conclusion Being active, regardless of weight classification (normal, overweight, or obese) was positively associated with cognitive function in this sample of older adults. This suggests that perhaps the best strategy to promote cognitive health in this population may be to encourage adequate levels of physical activity.