Abstract Aims Although loop diuretics are widely used to treat heart failure (HF), there is scarce contemporary data to guide diuretic adjustments in the outpatient setting. Methods and results In a prospective, randomized and double-blind protocol, we tested the safety and tolerability of withdrawing low-dose furosemide in stable HF outpatients at 11 HF clinics in Brazil. The trial had two blindly adjudicated co-primary outcomes: (i) symptoms assessment quantified as the area under the curve (AUC) of a dyspnoea score on a visual-analogue scale evaluated at 4 time-points (baseline, Day 15, Day 45, and Day 90) and (ii) the proportion of patients maintained without diuretic reuse during follow-up. We enrolled 188 patients (25% females; 59 ± 13 years old; left ventricular ejection fraction = 32 ± 8%) that were randomized to furosemide withdrawal (n = 95) or maintenance (n = 93). For the first co-primary endpoint, no significant difference in patients’ assessment of dyspnoea was observed in the comparison of furosemide withdrawal with continuous administration [median AUC 1875 (interquartile range, IQR 383–3360) and 1541 (IQR 474–3124), respectively; P = 0.94]. For the second co-primary endpoint, 70 patients (75.3%) in the withdrawal group and 77 patients (83.7%) in the maintenance group were free of furosemide reuse during follow-up (odds ratio for additional furosemide use with withdrawal 1.69, 95% confidence interval 0.82–3.49; P = 0.16). Heart failure-related events (hospitalizations, emergency room visits, and deaths) were infrequent and similar between groups (P = 1.0). Conclusions Diuretic withdrawal did not result in neither increased self-perception of dyspnoea nor increased need of furosemide reuse. Diuretic discontinuation may deserve consideration in stable outpatients with no signs of fluid retention receiving optimal medical therapy. ClinicalTrials.gov Identifier NCT02689180.