1. A prospective randomised trial was conducted to compare aminoglycoside dose prediction based on individually measured pharmacokinetic data, with dosage based on physician intuition. 2. After 2 days of therapy more patients in the pharmacokinetic group had achieved both peak (6-10 mg 1(-1] and trough (1-2 mg 1(-1] target plasma concentrations (P = 0.007), peaks alone (P = 0.01) and troughs alone (P = 0.01). Their mean (s.e. mean) peak concentration was 6.49 +/- 0.39 mg 1(-1) compared with 4.27 +/- 0.52 mg 1(-1) in the control group (P = 0.001), with trough concentrations of 1.44 +/- 0.22 mg 1(-1) and 0.94 +/- 0.21 mg 1(-1) respectively (P = 0.054). 3. After 5 days of therapy, peak and trough concentrations were still significantly higher in the pharmacokinetic group despite empirical dose adjustment (P = 0.01 and P = 0.013 respectively). 4. The mean (s.e. mean) daily dose of aminoglycoside was higher in the computer group (312 +/- 17 mg vs 203 +/- 13 mg, P = 0.001). 5. These findings suggest that dose estimation based on measured pharmacokinetic parameters is superior at achieving target plasma drug concentrations.