A randomized controlled trial was conducted to assess the efficacy of a computerized therapy programme for depression and anxiety, ‘Beating the Blues’, compared with cognitive-behaviour therapy provided by clinical psychologists and treatment as usual by general practitioners (GPs). The aim of the paper is to describe the study, the problems that were encountered when GPs agreed to recruit participants during consultations and to outline possible solutions to these problems. After three months only five participants had been recruited prompting modification of the study design. After one year only 17 patients had been recruited and the study was discontinued. The GPs indicated that the randomization procedure compromised their traditional role of providing patients with the best possible treatment. This produced role conflict that was resolved by GPs adhering to their care-giving role at the expense of recruiting to the study. GPs also felt that discussion of research took too much time during consultations and was inappropriate given that patients were often somewhat distressed. They also did not see the three treatment arms as equally acceptable, which resulted in GPs failing to recruit patients in case they were randomized to what they considered to be inferior treatments. Competing demands that rendered the study a low priority were also reported. We incorporated suggested solutions to these problems into the study which was rerun and completed after one year with 40 patients having been recruited.