The major objective of the Trimetazidine European Multicenter Study (TEMS) was to compare in a double-blind trial the anti-ischemic effects of trimetazidine (20 mg 3 times daily) with those of propranolol (40 mg 3 times daily). The inclusion criteria were based on an abnormal response to a multistage exercise test. After 3 months of treatment the improvements noted in all exercise testing data were similar in the trimetazidine and propranolol groups; similar data were obtained for the grades and severity of anginal attacks during daily life (from patient diaries). A 24-hour Holter monitoring was performed at entry and at the end of the study, but an abnormal Holter monitoring (1 -mm ST-segment depression during at least 1 minute) was not an inclusion criterion. This explains why at entry only 50% of the patients in both groups had an abnormal Holter recording. After 3 months of treatment, there were no significant differences between the 2 groups, but we observed a trend toward a decrease in ambulatory ischemia in the trimetazidine group and a trend toward an increase in ambulatory ischemia in the propranolol group. These data in the propranolol group are in total disagreement with the available literature on β blockers, which was due to a totally erratic behavior pattern in 2 patients in the propranolol group. When we excluded these 2 erratic cases from the propranolol group and extended our analysis to all available paired comparisons (day — 14 to day 30 and day 0 to day 90), we were able to compare 44 and 60 observations, both off therapy and on either propranolol or trimetazidine, respectively. The number of ischemic episodes was significantly reduced with trimetazidine (p <0.02) but not with propranolol. The total duration of ischemia was reduced (but not significantly) in the trimetazidine group; surprisingly, this reduction was essentially due to a reduction in symptomatic ischemic episodes. A circadian ischemic profile was noted in both groups, but it was notably more reduced in the trimetazidine group; surprisingly, this reduction was also essentially due to a reduction in symptomatic ischemic episodes. We conclude from these data that, (1) to draw conclusions on ischemic Holter monitoring data, an abnormal ischemic Holter should be one of the criteria to enter the study; (2) a 48-hour ambulatory recording is likely to provide more reliable information; and (3) much larger groups should be studied to avoid statistical interference from erratic cases.