Aminoglycosides are frequently used to treat sepsis in patients with liver disease. However, it has been suggested that cirrhotic patients are particularly sensitive to aminoglycoside-induced renal dysfunction. We investigated the efficacy and incidence of renal impairment with netilmicin plus mezlocillin compared with ceftazidime in 128 cirrhotic patients who required empirical treatment for sepsis. Renal impairment developed in 8 of 63 (13%) patients receiving netilmicin compared with 2 of 65 (3%) patients receiving ceftazidime (P < .05); it occurred despite regular monitoring of trough netilmicin levels. Renal impairment was present at the time of death in 1 of 13 (8%) patients treated with ceftazidime compared with 5 of 9 (56%) of the netilmicin patients (P < .05). Mortality rates were similar in the two groups (ceftazidime 20%, aminoglycoside 14%; P = NS). Renal dysfunction is significantly more frequent in cirrhotic patients treated with netilmicin but with careful attention to dosage and fluid management the clinical effect is likely to be relatively modest.