The mean and standard deviation of lignocaine (lidocaine) pharmacokinetic parameters in a patient population were determined on the basis of 327 serum concentration measurements obtained in 42 patients treated for ventricular arrhythmias. The application of a Bayesian forecasting method, which uses the estimates of the population parameters and 1 or 2 serum concentration measurements as feedback information, was tested retrospectively in 17 of the 42 patients (group I, 32 levels), and prospectively in 10 additional patients (group II, 20 levels). With I individual feedback concentration, sampled 2 to 4 hours after the start of lignocaine infusion, serum concentrations at 12 and 24 hours could be accurately predicted. The prediction error (measured minus predicted concentration) ranged between −1.2 and +1.6 (mean −0.03) mg/L in group I, and from −0.7 to +1.5 mg/L (mean +0.13) mg/L in group II; the correlation coefficient of measured and predicted levels were 0.92 and 0.86, respectively.In contrast, a prediction of lignocaine concentrations in these patients using only population parameters without feedback was poor: range of the prediction error = −3.1 to +3.0 mg/L (mean = +0.001 mg/L, r = 0.63, groups I and II, n = 52).The results demonstrate that with the Bayesian forecasting technique, accurate assessment of individual dosage requirements can be obtained within a few hours after starting lignocaine therapy.