Critically ill patients are frequently on drug treatment at the time of presentation,and usually receive additional drugs whilst in the intensive care unit. The necessity for each drug should be reviewed each day, risks weighed against benefits, and possible interactions borne in mind. Critically ill patients often have reduced functional reserves and adverse effects may be hard to detect. The use of a limited range of first-line drugs and of standard protocols developed with the aid of a designated pharmacist will minimize problems arising from inadequate stocks, incorrect storage, and errors in dispensing, dilution and administration. It has become fashionable to rationalize empirical dosing regimens on thebasis of systemic pharmacokinetic theory. However, this has contributed remarkably little to enhancing insight, or practices in the intensive care unit, as much of the available information is not relevant to critically ill patients, and because sudden changes in pathophysiology and blood concentrations cannot be accommodated by conventional pharmacokinetic analysis. There are also questions regarding the clinical relevance of drug blood concentrations, the influence of choice of site for blood sampling, the implications of the use of racemic mixtures, and of the effects of active metabolites. The use of physiological models and regional pharmacokinetics has the potential to provide the all-important 'missing link' between systemic pharmacokinetics and molecular pharmacology. However, much practical and theoretical work has yet to be done before this potential can be realized. For the moment, clinical practice should continue to be empirically based and reliance should be placed on the use of a few well-tried drugs and, whenever possible, on titrating dose against effect. © 1990 Baillière Tindall. All rights reserved.