Amikacin or gentamicin was used to treat 174 patients with suspected severe gram-negative infections in a prospective, randomized, double-blind trial. Enteric gram-negative bacilli were pathogenic in 71 cases (39 treated with amikacin, and 32 with gentamicin). Amikacin was effective in 10 to 12 bacteremias, 21 of 24 urinary-tract infections, two of five pneumonias and four of six other serious tissue infections. The toal favorable response rate was 77 per cent for amikacin and 78 per cent for gentamicin. Nephrotoxicity and auditory toxicity could be evaluated in 124 and 67 cases respectively. Definite nephrotoxicity developed in five of 62 (8 per cent) receiving amikacin and seven of 62 (11 per cent) given gentamicin, and possible nephrotoxicity developed in four patients in both groups. Definite ototoxicity developed in four patients in both groups. Definite otoxicity developed in two of 34 (6 per cent) and three of 30 (10 per cent) respectively. These differences were not statistically significant (by chisquare analysis, P greater than 0.05). The results indicate that amikacin is effective against severe gram-negative infections and is not more and not less ototoxic or nephrotoxic than gentamicin.