In heart cells, several distinct kinds of transient spatial patterns of cytoplasmic calcium ion concentration ([Ca2+]i) can be observed: (1) [Ca2+]i waves, in which regions of spontaneously increased [Ca2+]i propagate at high velocity (100 microns/s) through the cell; (2) Ca2+ 'sparks', which are spontaneous, non-propagating changes in [Ca2+]i that are localized in small (approximately 2 microns) subcellular regions; and (3) evoked [Ca2+]i transients that are elicited by electrical depolarization, in association with normal excitation-contraction (E-C) coupling. In confocal [Ca2+]i images, evoked [Ca2+]i transients appear to be nearly spatially uniform throughout the cell, except during their rising phase or during small depolarizations. In contrast to [Ca2+]i waves and spontaneous Ca2+ sparks, evoked [Ca2+]i transients are triggered by L-type Ca2+ channel current and they are 'controlled', in the sense that stopping the L-type Ca2+ current stops them. Despite their different characteristics, all three types of Ca2+ transient involve Ca(2+)-induced release of Ca2+ from the sarcoplasmic reticulum. Here, we address the question of how the autocatalytic process of Ca(2+)-induced Ca2+ release, which can easily be understood to underlie spontaneous regenerative ('uncontrolled'), propagating [Ca2+]i waves, might be 'harnessed', under other circumstances, to produce controlled changes in [Ca2+]i, as during normal excitation-contraction coupling, or changes in [Ca2+]i that do not propagate. We discuss our observations of Ca2+ waves, Ca2+ sparks and normal Ca2+ transients in heart cells and review our results on the 'gain' of Ca(2+)-induced Ca2+ release. We discuss a model involving Ca2+ microdomains beneath L-type Ca2+ channels, and clusters of Ca(2+)-activated Ca2+ release channels in the sarcoplasmic reticulum which may form the basis of the answer to this question.