Many of the important clinical decisions we make on a daily basis in stroke medicine are not supported by adequate evidence. This leads to variations in practice. If practice influences outcome, this must be regarded as unacceptable since it implies that many patients are receiving sub-optimal treatment. Where the advantages of certain treatment policies over others are only moderate, large randomised clinical trials provide the most reliable evidence of effectiveness. However, only a tiny proportion of patients with stroke are randomised in trials. Instead, the majority are exposed to treatments allocated haphazardly, rather than randomly, which serves only to delay the emergence of evidence concerning the relative merits of alternative treatment approaches. We suggest that we might increase the proportion of patients who contribute to advancing our knowledge by developing 'families' of trials. A 'family' would comprise a series of randomised trials into which patients with stroke may be enrolled either simultaneously or sequentially into one or more of the trials which would share common systems for randomisation and follow-up. Such a system would facilitate large, simple, randomised trials, reduce research costs, increase the generalisability of trial results and allow clinicians and patients to contribute to advancing our knowledge whenever they are uncertain about the best treatment. In this article, we discuss the advantages of this approach, some of the problems and their potential solutions.