Ca2+-activated aequorin luminescence and tension were measured in dog Purkinje fibres during twitches and during the increase in resting force produced by exposure of the fibres to a low-Na+ solution after 3 min without external K+. Over the restricted range which could be examined, the relation between tension and 'mean' aequorin luminescence (luminescence filtered at 0.2 Hz) was approximately linear during the development and maintenance of contracture. For a given level of force, the mean aequorin luminescence during contracture was up to 20 times greater than the peak luminescence during the twitch. Noise analysis of aequorin luminescence and tension during contracture indicated the presence of periodic fluctuations, with a predominant frequency in the range 1-4 Hz. Ryanodine (1 microM) or caffeine (10 mM) abolished the fluctuations in luminescence and tension and made the relation between tension and mean aequorin luminescence much steeper. A mathematical model, the key feature of which is periodicity in the asynchronous occurrence of spatially localized regions of relatively high [Ca2+], reproduces the experimental data derived from contractures. From the model analysis, we infer that tonic tension is produced by recruitment of increasing numbers of regions of high [Ca2+], rather than by homogeneous graded activation. These results indicate that during contracture or 'tonic tension', intracellular [Ca2+] is not at steady state, but rather undergoes large, asynchronous spatio-temporal fluctuations. Thus the assumptions that intracellular [Ca2+] is at steady state or homogeneous during tonic tension are not valid.