Aminoglycosides are usually given in two or three divided doses. A once-daily regimen might be more effective and less toxic. We have conducted a randomised trial in consecutive patients with serious infections for whom an aminoglycoside seemed warranted. Exclusion criteria were neutropenia or severely impaired renal function. 123 patients were enrolled. For efficacy analysis only those patients were considered in whom treatment with the aminoglycoside was not stopped within 72 h (n = 67); toxicity was analysed on patients receiving aminoglycosides for more than 48 h and not using other nephrotoxic medication (n = 85). Gentamicin 4 mg/kg every day (OD) or gentamicin 1.33 mg/kg three times daily (MD) (with dose-reduction in case of renal dysfunction) were given intravenously. In almost all patients intravenous amoxycillin 1 g every 6 h was also started. Baseline characteristics were comparable in both arms. A good clinical response was observed in 32/35 (91%) of the OD and in 25/32 (78%) in the MD group (difference 13%, 95% confidence interval -6.4% to +26.9%). 2 patients in each group died with uncontrolled infection. An insufficient bacteriological response (persistent positive cultures, resistance, or superinfection) was observed in 2 patients with OD and 3 patients with MD. In patients treated for more than 48 h duration of therapy and mean doses were 7.0 days (1590 mg) and 7.4 days (1672 mg) in OD and MD respectively. Mean first serum trough/peak levels were 0.6/10.2 mg/L and 1.4/5.2 mg/L. Nephrotoxicity (a rise in serum creatinine of 45 mumol/L or more) developed in 2/40 (5%) in OD and 11/45 (24%) in MD (p = 0.016). Risk factors for nephrotoxicity were duration of therapy and baseline creatinine clearance rate. High-tone audiometry was performed when possible; no significant differences were found in hearing loss (3/12 and 3/11) or prodromal signs of ototoxicity (5/12 and 4/11). A once-daily dosing regimen of gentamicin is at least as effective as and is less nephrotoxic than more frequent dosing.