Physical exercise testing and training in health and in heart disease may differ in concepts, targets and interpretation. Untrained males (n = 1962) of different age (18-65 years)--638 apparently healthy, 121 with arterial hypertension stage I-II, 1140 after confirmed acute myocardial infarction, and 63 after mitral surgery--were submitted to near-maximal bicycle spiro-ergometry (sitting, 60 rpm). Correlations between oxygen uptake (VO2), external work (W), heart rate (HR), systolic blood pressure (BP), double product (HR x BD = DP, as an indirect index of myocardial oxygen uptake), and other derived indices, varied substantially. The mutual predictive power of non-invasive cardiocirculatory parameters was weak particularly at low and moderate exercise (50-100 Watts). Different factors of priority influenced the variety in regression equations obtained in the separate groups. Discriminant values and limits rather than "norms" seem more appropriate in order to select the most relevant intervention (physical training, medical treatment, surgery). "Optimization" of disturbed inter-factor relations (e.g. of HR, BP, DP, W, VO2, VO2/kg, mechanical efficiency) may define the integral target of physical training in cardiac patients, if indicated. "Target-oriented" exercise prescription is discussed with examples. The cardiac cost of physical work (DP/VO2) is assessed in various heart conditions with its implications on exercise prescription. An attempt is made at differentiating between training the patient and "training" the diseased heart, and some controversies in terminology and in practical approach are stressed.