It is not easy to decide, when results from similar trials appear, whe
ther a trial still underway should be stopped or not. The weight of th
e other evidence has to be taken into account-as indeed it has to be f
or decisions by clinicians and by health service managers outside the
settings of a trial. Taking the Systolic Hypertension in the Elderly T
rial (SHEP) as an example, we show how its results are not unequivocal
(ie, there is no proof beyond reasonable doubt). This verdict justifi
es the continuation of similar trials in progress. More generally-but
again for individual clinicians and for trial organisers and again wit
h SHEP as the example-we illustrate a bayesian approach to trial-termi
nation decisions.