Both EVL and drug therapy are effective in the prevention of variceal rebleeding. Comparisons between the two modalities are few, and only in cirrhotics. This prospective randomized controlled trial compared EVL with drug therapy (propranolol + ISMN) in the prevention of rebleeds from esophageal varices in cirrhotic and noncirrhotic portal hypertension (NCPH) patients. One hundred thirty-seven variceal bleeders were randomized to EVL (Group I; n = 71) or drug therapy (Group II; n = 66). In Group I, EVL was done every 2 weeks till obliteration of varices. In Group II, propranolol (dose sufficient to reduce heart rate to 55 bpm/maximum tolerated dose) and ISMN (incremental dose up to 20 mg BD) were administered. Group I and II patients had comparable baseline characteristics, follow-up (12.4 vs. 11.1 months), cirrhotics and noncirrhotics [50(70.4%) and 21(29.6%) vs. 51(77.3%) and 15(22.7%)] and frequency of Child's A (35 vs. 27), B (26 vs. 28), and C (9 vs. 11). The mean daily dose was 109 +/- 46 mg propranolol and 34 +/- 11 mg ISMN and was comparable in cirrhotic and NCPH patients. Upper GI bleeds occurred in 10 patients in Group I (5 from esophageal varices) and in 18 patients in Group II (15 from esophageal varices) (P = 0.06). The actuarial probability of rebleeding from esophageal varices at 24 months was 22% in Group I and 37% in Group II (P = 0.02). The probability of bleed was significantly higher in Child's C compared to Child's A/B cirrhotics (P = 0.02). On subgroup analysis, in NCPH patients, the actuarial probability of bleed at 24 months was significantly lower in Group I compared to Group II (25% vs 37%; P = 0.01). In cirrhotics, there was no difference in the probability of rebleeding between patients in Group I and those in Group II (P = 0.74). In Group II, 25.7% patients had adverse effects of drug therapy and 9% patients had to stop propranolol due to serious adverse effects, none required stopping ISMN. There were 10 deaths, 6 in Group I (bleed related, 1) and 4 in Group II (bleed related, 1); the actuarial probability of survival was comparable (P = 0.39). EVL and combination therapy are equally effective in the prevention of variceal rebleeding in cirrhotic patients. EVL is more effective than drug therapy in the prevention of rebleeds in patients with NCPH and, hence, recommended. However, in view of the small number of NCPH patients, further studies are needed before this can be stated conclusively.