BACKGROUND: In randomized trials of primary prevention, there has been interest in the persistence of a beneficial effect of therapy after treatment has been stopped. We investigated the impact of measurement error in the outcome on results of the trial of preventing hypertension (TROPHY), a trial of candesartan vs. placebo for preventing hypertension in prehypertensive subjects. METHODS: We simulated the TROPHY study design, assuming that candesartan reduced blood pressure by 8/6 mm Hg during treatment, but had no carryover effect after treatment stopped. We simulated individual true blood pressures in the TROPHY-eligible range of 130-140 mm Hg for systolic blood pressure and 80-90 mm Hg for diastolic blood pressure, and added individual measurement variability. As in TROPHY, incident hypertension was defined as any three occurrences of systolic blood pressure 140 mm Hg or diastolic blood pressure 90 mm Hg. RESULTS: In the absence of any carryover effect, typical incidence curves for time to hypertension were similar to those from TROPHY. A significant difference in cumulative incidence 2 years after stopping treatment was detected in 80% simulated studies, giving a Type I error rate of 80%. CONCLUSIONS: The published data from TROPHY are consistent with a lack of carryover effect of candesartan.