Calcium sparks in frog intact skeletal muscle fibers were modeled as stereotypical events that arise from a constant efflux of Ca(2+) from a point source for a fixed period of time (e.g., 2.5 pA of Ca(2+) current for 4.6 ms; 18 degrees C). The model calculates the local changes in the concentrations of free Ca(2+) and of Ca(2+) bound to the major intrinsic myoplasmic Ca(2+) buffers (troponin, ATP, parvalbumin, and the SR Ca(2+) pump) and to the Ca(2+) indicator (fluo-3). A distinctive feature of the model is the inclusion of a binding reaction between fluo-3 and myoplasmic proteins, a process that strongly affects fluo-3's Ca(2+)-reaction kinetics, its apparent diffusion constant, and hence the morphology of sparks. DeltaF/F (the change in fluo-3's fluorescence divided by its resting fluorescence) was estimated from the calculated changes in fluo-3 convolved with the microscope point-spread function. To facilitate comparisons with measured sparks, noise and other sources of variability were included in a random repetitive fashion to generate a large number of simulated sparks that could be analyzed in the same way as the measured sparks. In the initial simulations, the binding of Ca(2+) to the two regulatory sites on troponin was assumed to follow identical and independent binding reactions. These simulations failed to accurately predict the falling phase of the measured sparks. A second set of simulations, which incorporated the idea of positive cooperativity in the binding of Ca(2+) to troponin, produced reasonable agreement with the measurements. Under the assumption that the single channel Ca(2+) current of a ryanodine receptor (RYR) is 0.5-2 pA, the results suggest that 1-5 active RYRs generate an average Ca(2+) spark in a frog intact muscle fiber.