Abstract Considering the increasing number of drugs evaluated for mania in randomised controlled trials (RCTs) and the potential discrepancies between recommendations based on RCTs and the antimanic treatment given in clinical practice, this paper addresses some issues related to RCTs on drug effects in mania. One major question raised in the paper is to what extent selection prior to the point of randomisation in RCTs in mania may limit the applicability of study results tto patients seen in ordinary clinical practice. Although such limitations in generalisability can be difficult to investigate empirically, it is emphasised that they should be openly discussed in the reports of RCTs. Another major focus is the issue of evaluation and interpretation of outcome, including a discussion of various response criteria based on mania rating scale scores. It is pointed out that essential criteria of dimensionality have only been sufficiently evaluated for the Bech-Rafaelsen Mania Rating Scale, although the fulfilment of such criteria are prerequisites for adding up the item scores to a total score reflecting the severity of mania. It is suggested that response defined as a decline in mania score below a certain limit may have some advantages over the commonly used 50 % reduction criterion. The issues arising from the unusual high drop-out rates of around 50 % are also addressed. Despite the fact that we need rigorous placebo-controlled trials to establish antimanic efficacy of new compounds, we also need large scale pragmatic studies using broad inclusion criteria, comparing the various treatments, alone or in combination, to investigate how they work in clinical practice. These studies may be randomised but open and use simple but relevant outcome measures.