Quantitative descriptions of the mechanical restitution curve as a description of variability in ventricular performance with coupling interval in isolated tissue preparations are widely available. In humans, however, in vivo examination of the force-interval relationship is restricted to test pulse intervals shorter than the sinus cycle length (i.e., incomplete mechanical restitution). The primary objectives in this investigation were therefore to examine this aspect of mechanical restitution in patients with ischemic heart disease and to develop a quantitative description of the phenomenon. Mechanical restitution curves were constructed in 40 patients, most of whom had well-preserved left ventricular (LV) systolic function, undergoing diagnostic cardiac catheterization for the investigation of chest pain, using a single premature test pulse interval during baseline atrial pacing. The mechanical restitution curve, the relationship between LV + dP/dtmax and test pulse interval, was fitted to a rectangular hyperbolic function. From this, the parameter c, the calculated proportional decrease in LV + dP/dtmax at 60% of the resting cycle length, was derived. The mechanical restitution curve-fitting model (involving determination of c) satisfactorily described the force-interval relationship in 37 of the 40 patients studied (as a rectangular hyperbola in 31 and with simple linear regression in 6 patients). The refractory period of the atria/atrioventricular node limited accurate use of the model in the remaining three patients. The parameter c was inversely proportional to both baseline atrial pacing cycle length (P < .001) and LV ejection fraction (P < .02) In patients with normal LV ejection fractions, the derived value of c at a cycle length of 800 ms (c800) was 29.0% baseline LV + dP/dtmax (95% confidence interval, 23.0, 35.0). The presence of hemodynamically significant ischemic heart disease was not a predictor of the parameters of the model. After intravenous injection of the beta-adrenoreceptor antagonist metoprolol in seven patients, there was a significant (P < .05) reduction in both c and LV + dP/dtmax at the baseline atrial pacing cycle length. Thus, the force-interval relationship can be quantitatively studied using incomplete mechanical restitution curves in humans in vivo. This quantitative description probably reflects relative intracellular calcium availability via slow channel activity and can be used to assess effects of cardioactive drugs on frequency-dependent inotropic mechanisms in humans. The predictive value of this mechanical restitution curve model for hemodynamic instability during tachycardia in patients with impaired LV function remains to be determined.