Statistical principles for analytical goal-setting were applied to two medical applications of drug-monitoring data: (a) individualizing dosage requirements by reference to a population-based therapeutic range or to a patient-specific decision value determined by Bayesian decision analysis, and (b) prospective dosing by using pharmaco-kinetic principles. For application a, the analytical goal for total allowable analytical error (TE) was defined as the amount of error that does not decrease by more than 5% either the sensitivity (probability) for detecting dosage regimens that may require modification or the specificity for detecting appropriately dosed patients. The limiting factor in achievable sensitivity and specificity was the intra-individual variation of peak steady-state concentration (Css), with the TE determined to be a CV of 4%. For application b, error-propagation rules were applied to a proposed prospective dosing scheme (J Pharmacokinet Biopharm 1978;6:135-51). The TE was determined to depend on the rate of theophylline clearance. For clearance rates less than or equal to 0.6 mL/(kg.min), the TE (CV) must not exceed 3% if the predicted infusion rate is to produce, with 95% confidence, a concentration within the therapeutic range.