The Frank-Starling relationship generally has been examined with filling pressure as the index of preload, resulting in a curvilinear function that plateaus at higher filling pressures. To investigate this relationship further in the intact heart, 32 dogs were chronically instrumented with left ventricular and pleural micromanometers and with regional (10 dogs) or global (22 dogs) ultrasonic dimension transducers. Seven days after implantation, left ventricular pressure and regional or global dimensions were recorded in the conscious state. After autonomic blockade, preload was varied by vena caval occlusion. Myocardial function was assessed by calculating regional or global stroke work, and preload was measured as end-diastolic segment length or chamber volume. The relationship between stroke work and either end-diastolic segment length or chamber volume (termed the preload recruitable stroke work relationship) was highly linear in every study (mean r = .97) and could be quantified by a slope (MW) and x-intercept (LW). Previous nonlinear relationships between stroke work and filling pressure seemed to reflect the exponential diastolic pressure-volume curve. Over the physiologic range of systolic arterial pressures produced by infusion of nitroprusside or phenylephrine, no significant change was observed in MW or LW in the normal dog. Calcium infusion increased both regional and global MW by 71 +/- 19% and 65 +/- 9%, respectively (p less than .02), with no significant change in LW. To normalize for ventricular geometry and heart rate, stroke work was computed from circumferential stress-strain data and converted to myocardial power output, which was then plotted against end-diastolic circumferential strain. This relationship also was highly linear, and the slope, Mmp (mW/cm3 of myocardium), is proposed as a potential measure of intrinsic myocardial performance independent of loading, geometry, and heart rate.