Background Treadmill exercise variables are powerful predictors of all-cause mortality but are unobtainable in at least 50% of patients because of disabilities precluding lower extremity exercise. Arm exercise stress testing is a potentially cost-effective alternative, but no long-term outcome data are available. Methods We performed arm ergometer stress tests on 446 veterans aged 64.0 (11.1) years (mean [SD]) between 1997 and 2002 and investigated whether arm exercise capacity in resting metabolic equivalents, heart rate recovery (in beats per minute), delta (peak resting) heart rate (in beats per minute), and other exercise variables predict long-term all-cause mortality, myocardial infarction (MI), or coronary revascularization. Results During follow-up of 12.0 (1.3) years, 255 patients died (57.2%), 70 had MI (15.7%), and 118 underwent coronary revascularization (26.4%). After adjustment for significant demographic and clinical variables, death was predicted by arm metabolic equivalents (hazard ratio/SD 0.59, 95% CI 0.46-0.75, P <.001), heart rate recovery (hazard ratio/SD 0.64, 95% CI 0.49-0.83, P <.001), and delta heart rate (hazard ratio/SD 0.75, 95% CI 0.63-0.91, P <.001). No exercise variables prognosticated MI, but coronary revascularization was predicted by stress-induced ST-segment deviations (hazard ratio 2.64, 95% CI 1.16-4.33, P <.001), limiting angina (hazard ratio 4.70, 95% CI 1.81-12.22, P <.001), and an abnormal perfusion imaging result (hazard ratio 2.0, 95% CI 1.14-3.51, P <.02). Conclusions Arm exercise capacity, heart rate recovery, and delta heart rate predict 12-year all-cause mortality and arm exercise-induced ST changes, limiting angina, and an abnormal nuclear imaging result portend coronary revascularization in lower extremity disabled veterans.