A substantial proportion of the psychiatric burden of disease falls on the world's poorest nations, yet relatively little is known about randomised trials conducted in these countries. Our aim was to identify and describe a representative sample of mental health trials from low and middle-income countries. 6107 electronic records, most with full text copies, were available following extensive searches for randomised or potentially randomised trials from low and middle-income countries published in 1991, 1995 and 2000. These records were searched to identify studies relevant to mental health. Data on study characteristics were extracted from the full text copies. Trials relevant to mental health were reported in only 3% of the records. 176 records reporting 177 trials were identified: 25 were published in 1991, 45 in 1995, and 106 in 2000. Participants from China were represented in 46% of trials described. 68% of trials had <100 participants. The method of sequence generation was described in less than 20% of reports and adequate concealment of allocation was described in only 12% of reports. Participants were most frequently adults with unipolar depression (36/177) or schizophrenia (36/177). 80% of studies evaluated pharmacological interventions, a third of which were not listed by WHO as essential drugs. 41% of reports were indexed on PubMed; this proportion decreased from 68% in 1991 to 32% in 2000. In terms of overall health burden, trial research activity from low and middle-income countries in mental health appears to be low, and in no area adequately reflects need.