In a cluster randomised clinical trial, all the participants who belong to the same cluster (eg, a local community, school, or general practice) are randomised to receive the same treatment. People from the same cluster tend to be more similar than those from different clusters, and the analysis must allow for this. A common enhancement is to add an assessment of participants' outcomes in a baseline period—that is, before randomisation. Even if different participants are assessed at baseline and follow-up, the fact that they are sampled from the same cluster allows some control for cluster differences. But the analysis of cluster randomised trials is more complex than for individually randomised trials, and it is not obvious how best to adjust for baseline assessments. A variety of methods of analysis have been suggested: in this article, we describe different approaches and their flaws, and make recommendations.