Multiple linear regression analyses were conducted on questionnaire data obtained from 33 agoraphobic women and their husbands before and after intensive exposure in vivo for the agoraphobia. The dependent variable in the regression equation was treatment outcome as determined by changes in composite ratings of patients' two main phobias. On this criterion, patients improved by 64% immediately after treatment and by 69% 1 year later. The main predictors of treatment outcome were patients' pre-treatment levels of agoraphobic disability and of extrapunitiveness. The main pre-treatment predictors of change during follow-up were patients' levels of social fear, and of fears which suggested dependency problems. When post-treatment questionnaire responses were used to predict patients' phobia change during follow-up, husbands' scores contributed 50% to the variance. This finding, together with clinical observations, suggested that if husbands were unable to fully acknowledge the emotional impact upon themselves of rapid and substantial improvements in their wives, then they were unlikely to facilitate further phobia improvement in their wives during the first 6 months of follow-up. It was concluded that involvement of the husbands in their wives' therapy would increase the likelihood of continuing phobia improvement after treatment.