Previous models of cardiac Ca2+ sparks have assumed that Ca2+ currents through the Ca2+ release units (CRUs) were similar to1-2 pA, producing sparks with peak fluorescence ratio <F/F-o> of similar to2.0 and a full-width at half maximum (FWHM) of similar to1 mum Here, we present actual Ca2+ sparks with peak F/F-o of >6 and a FWHM of similar to2 mum, and a mathematical model of such sparks, the main feature of which is a much larger underlying Ca2+ current. Assuming infinite reaction rates and no endogenous buffers, we obtain a lower bound of similar to 11 pA needed to generate a Ca2+ spark with FWHM of 2 mum. Under realistic conditions, the CRU current must be similar to 20 pA to generate a 2-mum Ca2+ spark. For currents greater than or equal to5 pA, the computed spark amplitudes (F/F-o) are large (similar to6-12 depending on buffer model). We considered several factors that might produce sparks with FWHM similar to2 mum without using large currents. Possible protein-dye interactions increased the FWHM slightly. Hypothetical Ca2+ "quarks" had little effect, as did blurring of sparks by the confocal microscope. A clusters of CRUs, each producing 10 pA simultaneously, can produce sparks with FWHM similar to2 mum. We conclude that cardiac Ca2+ sparks are significantly larger in peak amplitude than previously thought, that such large Ca2+ sparks are consistent with the measured FWHM of similar to2 mum, and that the underlying Ca2+ current is in the range of 10-20 pA.