1. The influence of caffeine, applied over a 25-fold range of concentrations, on intramembrane charge movements was examined in intact voltage-clamped amphibian muscle fibres studied in the hypertonic gluconate-containing solutions that were hitherto reported to emphasize the features of qgamma at the expense of those of qbeta charge. 2. The total charge, Qmax, the transition voltage, V*, and the steepness factor, k, of the steady-state charge-voltage relationships, Q(V), were all conserved to values expected with significant contributions from the steeply voltage-dependent qgamma species (Qmax approximately 20 nC microF-1, V* approximately -50 mV, k approximately 8 mV) through all the applications of caffeine concentrations between 0.2 and 5.0 mM. This differs from recent reports from studies in cut as opposed to intact fibres. 3. The delayed transients that have been attributed to transitions within the qgamma charge persisted at low (0.2 mM) and intermediate (1.0 mM) caffeine concentrations. 4. In contrast, the time courses of such qgamma currents became more rapid and their waveforms consequently merged with the earlier qbeta decays at higher (5.0 mM) reagent concentrations. The charging records became single monotonic decays from which individual contributions could not be distinguished. This suggests that caffeine modified the kinetic properties of the qgamma system but preserved its steady-state properties. These findings thus differ from earlier reports that high caffeine concentrations enhanced the prominence of delayed transient components in cut fibres. 5. Caffeine (5.0 mM) and ryanodine (0.1 mM) exerted antagonistic actions upon qgamma charge movements. The addition of caffeine restored the delayed time courses that were lost in ryanodine-containing solutions, reversed the shift these produced in the steady-state charge-voltage relationship but preserved both the maximum charge, Qmax, and the steepness, k, of the steady-state Q(V) relationships. 6. Caffeine also antagonized the actions of tetracaine on the total available qgamma charge, but did so only at the low and not at the high applied concentrations. Thus, 0.2 mM caffeine restored the steady-state qgamma charge, the steepness of the overall Q(V) function and the appearance of delayed qgamma charge movements that had been previously abolished by the addition of 2.0 mM tetracaine. 7. In contrast, the higher applied (1.0 and 5.0 mM) caffeine concentrations paradoxically did not modify these actions of tetracaine. The total charge and voltage dependence of the Q(V) curves, and the amplitude and time course of charge movements remained at the reduced values expected for the tetracaine-resistant qbeta charge. 8. These results permit a scheme in which caffeine acts directly upon ryanodine receptor (RyR)-Ca2+ release channels whose consequent activation then dissociates them from the tubular dihydropyridine receptor (DHPR) voltage sensors that produce qgamma charge movement, with which they normally are coupled in reciprocal allosteric contact.