The literature reviewed herein clearly demonstrates the poor correlation between drug dosing and the ability to achieve a specific serum drug concentration and between drug dosing and clinical response, especially for drugs with a narrow therapeutic index. There is, however, a better correlation between serum drug concentration and observed clinical response. Thus, clinicians use serum drug concentrations to more accurately dose drugs. Numerous dosing methods have been developed in an attempt to improve the relationship between dosing, serum drug concentration, and response. The major hypothesis is that if dosing methods can be developed that will accurately predict serum drug concentrations, these methods would be useful in improving clinical care. Several dosing methods have been developed including use of 'standard' doses, population-based predictive algorithms and nomograms, pharmacokinetic equations, and Bayesian feedback. Some of these methods are accurate and useful, whereas others are not. This review evaluates the commonly used dosing methods (some of which utilise serum drug concentration feedback for dosage estimation) for 5 drugs: gentamicin, digoxin, phenytoin, theophylline, and lignocaine (lidocaine). These drugs were selected since they exhibit a representative cross section of pharmacokinetic parameters and since they exhibit a representative cross section of pharmacokinetic parameters and since they have narrow therapeutic ranges. An individualised method and a Bayesian method, both using serum drug concentration feedback, appear most accurate and precise in dosing to achieve desired serum drug concentrations and, hence, response. Our bias from personal experience with this method and from published use by others is that the Bayesian method is more flexible in that any number of serum drug concentrations may be used to determine dose, instead of the 3 or more required for the individualised method. Although use of these methods would appear to be cost-effective in timely provision of health care by reduction of toxicity and hospital stay, only sparse data have been generated to support this conclusion. Thus, further examination of the cost-effectiveness of drug dosing methods is necessary to establish their place in routine patient care.