Forty-nine dogs showing signs of separation-related problems were randomly assigned to one of three groups: group A (15 dogs) received a placebo twice daily; group B (17 dogs) received clomipramine at 0.5 to 1.0 mg/kg twice daily; and group C (17 dogs) received clomipramine at 1.0 to 2.0 mg/kg twice daily. All the dogs also received behavioural therapy. Their owners were required to complete questionnaires about their dog's behaviour initially, and one, four and eight weeks after the treatment with clomipramine began. Bipolar ratings scales were used to monitor the frequencies of 'general', 'attachment-related' and 'separation-related' behaviours. Kruskal-Wallis tests and Kendall Rank correlations were used to determine any initial differences between the treatment groups, and the association between the initial scores and behavioural changes after one week of treatment with clomipramine. Extended Mantel-Haenszel statistics were used to evaluate the effects of clomipramine treatment versus the placebo, and Page's test was used to assess the effectiveness of behavioural therapy on its own. There were no significant differences in the demographic characteristics of the owners of the dogs assigned to the three groups. The dogs differed slightly in age between groups, and the dogs in the two clomipramine-treated groups were reported as showing problems at a significantly earlier age than those in the placebo group. Clomipramine treatment had a sustained suppressive effect on the dogs' general activity levels, and a more modest suppressive effect on their attachment-related tendency to want much physical contact with their owners. The typical signs of separation-related behaviour problems were not significantly affected by treatment with clomipramine, but behavioural therapy on its own was highly effective in reducing behavioural problems.