To assess the feasibility of diuretic discontinuation in patients with stable congestive heart failure (CHF) and to identify risk factors for subsequent development of congestion, a prospective, 12-week clinical trial of unmasked diuretic withdrawal was conducted with continuation of background CHF therapy and double-blind randomization to placebo or lisinopril. Forty-one patients with a history of CHF and continuous diuretic use for > or = 3 months had all diuretic therapy discontinued, and therapy with lisinopril 5 mg (target 20 mg)/day (n = 20) or placebo (n = 21) begun the next day. A diuretic was restarted if new or worsening CHF symptoms and signs developed. Twelve patients (29%) did not require diuretic reinitiation at any time during follow-up, whereas 29 (71%) restarted diuretic therapy after a median of 15 days (range 2 to 42). Fourteen patients taking lisinopril and 15 taking placebo required diuretic drugs (p = NS). The baseline daily furosemide dose of > 40 mg, a left ventricular ejection fraction < or = 0.27, and history of systemic hypertension were independently predictive of early diuretic reinitiation by Cox proportional-hazards analysis. The probability of remaining diuretic-free after 6 weeks was 71% if none of these criteria were present. This trial demonstrates the feasibility of discontinuing diuretic drugs in certain patients with stable CHF and predicts those patients likely to require reinitiation of therapy. Diuretic withdrawal may be warranted when the furosemide dose is < or = 40 mg/day, left ventricular ejection fraction is > 0.27 and when no history of systemic hypertension is present.