The possibility of using alteration in the ventricular gradient (VG) to discern the presence of ischemic heart disease was studied in 30 patients with effort angina pectoris(AP), 21 with vasospastic angina (VSA), 21 with chest pain syndrome (CPS), and 20 healthy volunteers (control). The VG of each consecutive heart beat over a 22-sec interval was calculated by microcomputer from resting Frank-lead X, Y, Z scalar electrocardiograms. The mean values and standard deviations (SD) for the azimuth, elevation, and magnitude of the VG in each group were calculated. The SD and SD/mean ratios for each parameter were used as indices of VG alteration in the groups, and the indices were compared. The SD and SD/mean for the magnitude and elevation of VG were significantly greater in the AP group than in the CPS and control groups. The SD for the azimuth of VG was significantly greater in the AP group than in the CPS and control groups (p < 0.01). The SD and SD/mean for the magnitude of VG were greater in the AP group than in the VSA group (p < 0.01). The SD/mean for the elevation and magnitude of the VG were greater in the VSA group than in the control group (p < 0.01). The SD/mean of the magnitude of the VG was found to be the best index, as it was higher than the upper limit of the control group. The sensitivity and specificity were 80 and 91% (AP vs CPS, p < 0.001), and 43 and 91% (VSA vs CPS, not significant), respectively.(ABSTRACT TRUNCATED AT 250 WORDS)