Background Resting heart rate is directly associated and maximal exercise-induced heart rate inversely associated with cardiovascular mortality, and therefore their difference might contain prognostic information from both variables. The comparative long-term prognostic values of maximal exercise-induced heart rate and of the difference between it and resting heart rate were studied in apparently healthy middle-aged men. Methods Resting heart rate and maximal exercise-induced heart rate were measured, and their difference calculated, in 1960 apparently healthy men aged 40–59 years, and mortality was recorded over a period of 16 years. Conventional coronary risk factors were assessed at baseline. Results Both the difference between the two heart rates and the maximal exercise-induced heart rate were strongly, independently and inversely associated with cardiovascular mortality after adjustment for age, smoking, systolic blood pressure, lung function, glucose tolerance, serum cholesterol level, serum triglycerides level, physical fitness and exercise ECG findings. The adjusted relative risk of cardiovascular death in heart-rate difference quartiles 3 and 4 compared with that in quartile 1 (the lowest heart-rate difference quartile) was 0.54 (95% confidence interval 0.33–0.86; P= 0.009). The corresponding value for maximal exercise-induced heart rate was 0.56 (95% confidence interval 0.34–0.89; P=0.018). Within the lowest heart-rate difference quartile, but not within the lowest maximal exercise-induced heart rate quartile, a further, strong, negative gradient in cardiovascular mortality was observed. In the high working capacity range, low heart-rate difference but not low maximal exercise-induced heart rate predicted very high cardiovascular disease mortality. Heart-rate difference and maximal exercise-induced heart rate were also inversely associated with non-cardiovascular disease mortality. Conclusions Both heart-rate difference and maximal exercise-induced heart rate were strong, graded, long-term predictors of cardiovascular mortality among apparently healthy middle-aged men, independent of age, physical fitness and conventional coronary risk factors. However, low heart-rate difference was a better predictor than low maximal exercise-induced heart rate for recognizing individuals who were at particularly high risk of dying prematurely from cardiovascular diseases.