All therapeutic decisions involve some trade-off between therapeutic benefits and risks; a new therapy may be associated with greater efficacy but also a greater risk of adverse effects. In making treatment decisions clinicians must examine the clinical evidence regarding the magnitudes of benefit and risk and the precision with which they have been estimated. Ideally this requires a systemic assessment of the quality of the research and the strength of the evidence. We examine how the concept of number needed to treat can be used to improve the current presentation of clinical trials data of efficacy and side-effects to give clinicians a more clinically meaningful and quantitative measure of benefit-risk trade-offs. We propose a benefit-risk ratio that quantifies for a new therapy how many therapeutic (efficacy) events will be achieved for each adverse event incurred. We show how data from a clinical trial with a single binary measure of efficacy and a single adverse event of concern can be used to provide point estimates and confidence intervals for the benefit-risk ratio. The approach is illustrated using data from the GUSTO trial comparing tissue plasminogen activator and streptokinase in the management of patients with acute myocardial infarction.