The application of clinical pharmacological concepts and therapeutic standards in intensive care settings presents particularly difficult problems due to the lack of adequately controlled background information and the highly variable and rapidly evolving clinical conditions where drugs must be administered and their impact evaluated. In this review, an attempt has been made to discuss the available knowledge within the framework of a problem-oriented approach, which appears to provide a more clinically useful insight than a drug-centred review. Following a brief discussion of the scanty data and the most interesting models to which reference can be made from a pharmacokinetic point of view (the burn patient being taken as an example), the review concentrates on the main general intervention strategies in intensive care patients. These are based mainly on non-pharmacological measures (correction of fluid and electrolyte balance, total parenteral nutrition, enteral nutrition, oxygenation and ventilatory management) and are discussed with respect to the specific challenge they present in various clinical conditions and organ failure situations. In addition, 4 major selected clinical conditions where general management criteria and careful use of prophylactic and therapeutic drug treatments must interact to cope with the variety of presentations and problems are reviewed. These include: acute cerebral damage; anti-infective prophylaxis and therapy; cardiovascular emergencies; and problems of haemostasis. Each problem is analysed in such a way as to frame the pharmacological intervention in its broader context of the underlying (established or hypothesised) pathophysiology, with special attention being paid to those methodological issues which allow an appreciation of the degree of reliability of the data and the recommendations which appear to be practiced (often haphazardly) in intensive care units. The thorough review of the published literature provided (up to mid-1986) clearly shows that in this field the quality of randomised controlled and epidemiological studies is rather unsatisfactory. It would be highly beneficial to research and to clinical care if larger multicentric protocols and prospective epidemiological comparative investigations could be carried out to investigate more timely and adequately the variables which determine drug action, and the final outcome in the many subgroups of patients which must be considered in a proper stratification of intensive care unit populations.