Intramembrane charge movement and myoplasmic free calcium transients (delta[Ca2+]) were monitored in voltage-clamped segments of isolated frog muscle fibres cut at both ends and mounted in a double Vaseline-gap chamber. The fibres were stretched to sarcomere lengths of 3.5-4.6 micron to minimize mechanical movement and the related optical artifacts. The over-all calcium removal capability of each fibre was characterized by analysing the decay of delta[Ca2+] following pulses of several different amplitudes and durations. The rate of sarcoplasmic reticulum (s.r.) calcium release was then calculated for each delta[Ca2+] using the calcium removal properties determined for that fibre. The calculated calcium release wave form reached a relatively early peak and then declined appreciably during a 100-150 ms depolarizing pulse. The voltage dependence of the peak rate of calcium release was steeper and was centred at more positive membrane potentials than the steady-state voltage dependence of charge movement in the same fibres. A considerable fraction of the total intramembrane charge was moved at potentials at which delta[Ca2+] and calcium release were only a few per cent of maximum. This 'subthreshold' charge may correspond to charge moved in preliminary transitions that precede a final charge transition that activates release. A 'stepped on' pulse protocol was used to experimentally separate the subthreshold charge movement from the charge movement of the final transitions that may control calcium release. The stepped on pulse consisted of a set 50 ms pre-pulse to a potential just at or below the potential for detectable delta[Ca2+] followed immediately by a test pulse of varying amplitude and duration. For a wide range of test pulse amplitudes and durations in the stepped on protocol the peak rate of calcium release was linearly related to the charge movement during the test pulse. This result points to a tight control of activation of s.r. calcium release by intramembrane charge movement. The voltage dependence of both charge movement and of the rate of calcium release could be fitted simultaneously with a three-state, two-transition sequential model in which charge moves in both transitions but only the final transition activates s.r. calcium release. A model with three identical and independent charged gating particles per channel gave an equally good fit to the data. Both models closely fit the charge movement and release data except within about 10 mV of the voltage at which release became detectable, where release varied more steeply with membrane potential than predicted by either model.(ABSTRACT TRUNCATED AT 400 WORDS)