Abstract Background and aims To investigate if frequency of outdoor recreational activity (ORA) predicts cardiovascular disease (CVD) mortality, independent of serum 25(OH)D concentration. Methods and results Baseline data on ORA and serum 25(OH)D, collected from 11,746 participants aged 30–90 years in the Third National Health and Nutrition Examination Survey during 1988–1994, were linked to the National Death Index for assessment of CVD deaths from baseline through December 2006. CVD mortality as a primary cause of death was assessed during a mean follow up of 12.9 (SD, 4.2) years. There were 1519 CVD deaths during follow up. A strong positive association was observed between frequency of ORA in the last month and serum 25(OH)D (p < 0.001). Compared to participants who did no ORA in the last month, the hazard ratio (HR) of CVD mortality was 0.72 (95% confidence interval 0.58–0.90) for those doing ORA 1–4 times, 0.64 (0.47–0.89) for 5–12 times, 0.70 (0.56–0.89) for 13–30 times and 0.63 (0.47–0.84) for ≥30 times (p-trend < 0.001), in a Cox proportional hazards regression model which included 25(OH)D and CVD risk factors. Serum 25(OH)D was inversely associated with CVD mortality (p-trend, 0.01) in this same model. Conclusions An inverse association between ORA and CVD mortality was observed independent of 25(OH)D. The underlying mechanism for this association may not involve 25(OH)D hence, further studies are warranted to confirm and investigate the underlying mechanism.