The maximum amplitude of vector loops formed by summing orthogonally recorded bipolar electrograms has been shown to reflect the direction of activation in cardiac muscle. To investigate whether components of vector loops could provide information about different activation directions in local areas of myocardium, we correlated "instantaneous vectors" with isochronal activation patterns in an in vitro preparation of experimental myocardial infarction. In thirteen 3 mm x 3 mm regions studied (from 11 tissues), at least 16 microelectrode impalements with a minimum density of 0.8 mm between sites were made. In seven situations in which notched, irregular, and prolonged duration electrograms were present, vector loops pointed in the same general direction throughout their entire time course. In these preparations, microelectrode impalements demonstrated only a single major direction of activation. In five of six areas in which multidirectional vector loops were present, two or more separate local directions of activation corresponded to the directions of the vector loop. Instantaneous vectors were then used to analyze propagation patterns in vivo in 10 animals with 2-4-week-old experimental myocardial infarction. Of 150 sites in the 10 animals, 13% contained more than one major local direction of activation. In 11 markedly abnormal sites, electrograms were recorded during pacing from four sites around the recording probe. When comparing electrogram characteristics from the four sites, a mean difference of 5.9 mV in electrogram amplitude and 19.1 msec in electrogram duration (coefficient of variation, 27% for amplitude and 22% for duration) was found. In only two of the 11 sites was it found that the same number of activation directions occurred from all pacing sites. We conclude: 1) Instantaneous components of vector loops accurately represent local directions of cardiac activation at differing times. 2) Most areas of experimental myocardial infarction have only one major direction of activation despite the presence of abnormal electrograms. 3) In some regions, however, two major local directions of activation can be identified within a relatively local area. 4) Geometric activation patterns in experimental myocardial infarction are markedly dependent on initial activation direction.