In isolated rat left ventricles perfused at 37 degrees C with control, Ca2+-loading, and Ca2+-depleting solutions (pH 7.3-7.4), we have investigated freeze-fractured gap junctional membrane by three quantitative techniques designed to correlate changes in junctional permeability with changes in membrane ultrastructure, i.e., 1) optical diffraction, 2) direct measurement of center-to-center spacings and particle diameters, and 3) statistical analysis of the spatial distribution of P-face particles based on analysis of nearest neighbor center-to-center distances. Junctions fixed either with glutaraldehyde or by quick freezing were compact, with closely packed rather than dispersed membrane particles even in the permeable state. Analysis of variance for all three methods indicated that replication was a major variability source limiting structural discrimination. Discrimination between random, regular, and clustered distributions depended critically on particle diameter and particle density. The results differ from published data of others on mammalian ventricular gap junctions and from measurements by our laboratory on sheep cardiac Purkinje fibers (J. Ultrastruct. Res. 75: 195-204, 1981).