The cornerstone of evidence-based medicine is the belief that good quality research should form the basis of clinical practice and decision-making (Muir Gray, 1997). Psychiatry has kept abreast of this movement (Geddes et al. 1997) and claims have been made that randomized-controlled trials (the highest quality primary evaluative research) can be used to justify 65% of routine clinical decisions (Geddes et al. 1996). However, it is largely published research that forms the ‘knowledge base’ of the evidence movement. A fundamental difficulty arises when published research results are a biased sample of all research results – published and unpublished. Publication bias presents one such threat and has been much discussed in wider healthcare (Easterbrook et al. 1991; Dickersin & Min, 1993; Dickersin, 1997), but has been little discussed or researched in psychiatry, despite the fact that psychiatry is likely to be at least as prone to publication bias as other specialities.