BACKGROUND We set out to investigate whether community nurses could be trained in problem-solving therapy and, once trained, how effective they would be in treating emotional disorders in primary care.

METHOD Seventy patients with an emotional disorder in primary care were randomly allocated to receive either problem-solving therapy from a trained community nurse or treatment as usual from their general practitioner. Interview and self-rated assessments of clinical and economic outcome were made pretreatment, at eight weeks and at 26 weeks after treatment.

RESULTS There was no difference in clinical outcome between patients who received problem-solving treatment and patients who received the general practitioner's usual treatment. However, patients who received problem-solving treatment had fewer disability days and fewer days off work. The health care cost of problem-solving was greater than that of the general practitioner's usual treatment but this was more than offset by savings in the cost of days off work.

CONCLUSIONS Problem-solving treatment can be given by trained community nurses. The clinical effectiveness and cost-benefit of the treatment will depend on the selection of appropriate patients.