The pathophysiology of sepsis has been studied intensively in recent years and a variety of opportunities for therapeutic intervention have been identified. A number of biological products including endotoxin antibodies, cytokine inhibitors and receptor antagonists have been evaluated after the failure of pharmacological doses of steroids to influence survival in septic shock. Despite a number of large, international multi-centre studies, the therapeutic promise of these various interventions remains unfulfilled. These trials have largely been conducted in intensive care units in a heterogeneous population of patients with various entry criteria and end-points of response. While the clinical trial must remain the standard for assessing safety and efficacy of new interventions there are opportunities to improve on the design, execution and analysis of these studies. Factors such as the appropriateness of antibiotic therapy, the adequacy of medical and surgical management, and the issue of withdrawal or withholding of life support are discussed in relation to these studies. Furthermore the role of an independent scientific extramural review committee is stressed, particularly in relation to the impact of confounding events of an unforeseen nature. The potential for improving the quality of the analyses of clinical trials of sepsis is illustrated by a recently completed study of the efficacy of a murine monoclonal antibody to human tumour necrosis factor-alpha.