SUMMARY1. Simultaneous measurements were made in anaesthetized dogs of monophasic action potentials from the right ventricle and of the maximum rate of rise of left ventricular pressure (dPiv/dtmax). Atrio-ventricular dissociation was induced and the heart paced via right ventricular electrodes.2. A control period of steady pacing was followed by a test stimulus after a variable interval called the 'test-pulse interval'. The duration of the action potential of the test beat (measured at 70 % repolarization) increased with test-pulse interval and reached an approximately steady value at intervals of 10-1-5 sec. This constitutes the electrical restitution curve'.3. An increase in the frequency of stimulation prior to the introduction of the test pulses caused a downward displacement of the electrical restitution curve.4. Stimulation at 2 Hz and paired pulse stimulation at 1 Hz (same number of stimuli per min) were introduced prior to the test pulses and produced very similar electrical restitution curves.5. For a constant frequency of stimulation in the control period, adrenaline produced downward displacement of the restitution curve.6. It is concluded that there is no obvious relationship between the restitution of the action potential duration and of the contractile response. We suggest therefore that electrical and mechanical restitution occur through separate processes, the former through time-dependent recovery in membrane conductances and the latter through time-dependent increase in availability of intracellular calcium for release.7. Contractions were introduced with a test-pulse interval shorter than the optimum, and were followed by a second test pulse fixed at the optimum interval of 0-8-1 0 sec. The second test beats were potentiated (post-extrasystolic potentiation). In isolated ejecting cat hearts, there was an optimum interval for the first test pulse * Address for correspondence. G. ELZINGA AND OTHERS to produce the greatest potentiation of the second test beat. This interval was 0-2-0{3 sec, and was shortened by an increase in frequency of stimulation prior to the first test beat.8. The interval preceding the first test pulse was then varied within a range (0-8-2-0 sec) which did not produce potentiation. These first test pulses were sometimes preceded by one extrasystole. The timing of this extrasystole was altered to vary the post-extrastolic potentiation of the first test pulse.9. Multiple regression analysis, carried out between dPiv/dtmax of the second test pulse (DP2, the dependent variable) and the action potential duration (AP1) and dPiv/dtmax (DP1) of the first test pulse (independent variables) yielded correlation coefficients between 0x88 and 0 99. Each determination of the coefficient included data from beats with and without post-extrasystolic potentiation.10. It is postulated that the coefficient relating DP2 to DP1 in the multiple regression analysis (mean value 0 75) is an index of the proportion of calcium stored during relaxation which is released again on the next beat.11. When...