Short-term (4-8 weeks) placebo-controlled trials are used to evaluate new antihypertensive drug treatment. To evaluate the consequences of such practice, a descriptive meta-analysis was conducted, consisting of blinded review of original case report forms for all patients who died or left a study before its completion for all short-term, placebo-controlled hypertension trials submitted to the Food and Drug Administration from 1973 through 2001. There were 93 marketing applications or supplements involving 590 individual trials that involved 86 137 randomized patients (64 438 randomized to experimental drug and 21 699 randomized to placebo) with 12 658 patient years of observation. There were 9636 dropouts (mean time to dropout was 28 days) and relative risk (RR (placebo/drug)) ¼ 1.33 (95% confidence limits, 1.28, 1.39; Po10 À16 ). As expected, lack of blood pressure (BP) control was far more common in patients randomized to placebo; therapeutic failure, RR ¼ 2.53 (2.35, 2.73; Po10 À15 ) and hypertensive emergency, RR ¼ 2.75 (2.19, 3.57; Po10 À15 ). When administrative dropouts and dropouts resulting from inadequate BP control were excluded, the remaining 38% of dropouts were disproportionately more from drug (2810 drug, 816 placebo), RR ¼ 0.80 (0.74, 0.86; Po10 À8 ). There were 43 deaths, RR ¼ 0.72 (0.33, 1.45; P ¼ 0.37); 40 strokes, RR ¼ 1.43 (0.68, 2.81; P ¼ 0.33) and 77 myocardial infarctions, RR ¼ 1.06 (0.62, 1.75; P ¼ 0.82). Irreversible harm (a combination of death, stroke and myocardial infarction, 160 total events) was equally distributed between the drug and placebo groups, RR ¼ 1.03 (0.71, 1.47; P ¼ 0.86).