Leisure-time physical activity (LTPA) is associated with longevity in patients with coronary artery disease (CAD). However, less is known about prognostic significance of longitudinally assessed LTPA in patients with stable CAD. The present study assessed the relationship between changes in LTPA and cardiac mortality in patients with CAD. Patients with angiographically documented CAD (n = 1,746) underwent clinical examination and echocardiography at the baseline. Lifestyle factors, including LTPA (inactive, irregularly active, active, highly active), were surveyed at baseline and after 2 years' follow-up. Thereafter, the patients entered the follow-up (median: 4.5 years; first to third quartile: 3.4 to 5.8 years) during which cardiac deaths were registered (n = 68, 3.9%). The patients who remained inactive (n = 114, 18 events, 16%) and became inactive (n = 228, 18 events, 8%) had 7.6- (95% confidence interval [CI] 4.2 to 13.6) and 3.7-fold (95% CI 2.1 to 6.7) univariate risk for cardiac death compared with those who remained at least irregularly active (n = 1,351, 30 events, 2%), respectively. After adjustment for age, gender, body mass index, diabetes, previous myocardial infarction, left ventricular ejection fraction, angina pectoris grading, cardiovascular event during initial 2-year follow-up, smoking and alcohol consumption, the patients who remained inactive and became inactive still had 4.9- (95% CI 2.4 to 9.8, p <0.001) and 2.4-fold (95% CI 1.3 to 4.5, p <0.01) risk for cardiac death, respectively, compared with patients remaining at least irregularly active. In conclusion, LTPA has important prognostic value for cardiac death in patients with stable CAD. Even minor changes in LTPA over 2 years were related to the subsequent risk for cardiac death.