Objectives Using an evidence-based approach for a round table conference, to discuss sepsis according to its current epidemiology and clinical management, lessons which we feel can be learned by investigators from the design and conduct of previous clinical trials of drug therapy, and to describe the ``optimum'' clinical trials design for treatments of this syndrome. Data Sources Experts in the field of sepsis were selected and requested to apply an ``evidence-based'' approach to the published literature, from which recommendations would be synthesized and rated according to levels of evidence. These experts undertook a review of appropriate literature, primarily focusing on evidence presented by clinical trials. Applicable articles were searched for by individual experts, using a variety of on-line search strategies (i.e., MEDLINE, Current Contents; Clinical Medicine). Data Extraction and Review Presentation at the round table conference was followed by agreement by consensus amongst participants as to the grade of recommendation. Individual presentations and arguments for grading of levels of evidence will be published independently. Where possible, recommendations on individual topics were ranked according to the level of evidence presented. Levels I to V were used to rank randomized, controlled trials through case series, respectively. Grade A recommendation is supported by Level I evidence, Grade B recommendation is supported by Level II evidence, and Grade C recommendation is supported by Levels III, IV or V evidence. Conclusion Recommendations for the design, conduct and analysis of future trials of sepsis therapies were summarized. (Crit Care Med 1995; 23:394-399)