An investigation of marital distress was divided into three separate but related studies. The first involved the investigation of the behavioural model of marital distress. Twenty distressed and twenty nondistressed couples were compared using a multi-dimensional battery of measures, including self-report, quasi-observational, and laboratory observational measures. Several multivariate analyses were performed on the data. In general, results supported the behavioural model which describes marital distress as a failure of partners to exchange a sufficiently high number of Pleases relative to Displeases, a failure of partners to resolve conflict, and a subsequent development of avoidance behaviours. Consistent with previous results, a stronger relationship was found between marital distress and negative rather than positive behaviours, and the conflict score, as measured by the Areas of Change Questionnaire (ACQ) was found to be the most powerful discriminating of all the self-report and quasi-observational variables. Results of discriminant analyses suggested that husbands' marital satisfaction was more strongly influenced by earlier relationships than was wives' satisfaction, and that wives more readily translated behaviours occurring within a distressed relationship into subjective feelings. The second study investigated the treatment of marital distress using a structured behavioural-based group training programme. Treatment gains were evaluated using a similar but expanded multi-dimensional battery of measures pre- and posttreatment, and maintenance of gains was evaluated by means of repeated measures on the full test battery at 6, 12, and 18 months, and on an abbreviated version at 3, 9, and 15 months posttreatment. Thirteen couples were trained, and seven couples continued to provide data for the full 18 months. Data were both pooled and investigated as a series of single cases. Strong treatment effects (p <.001) were demonstrated when multiple analyses of variance were performed on pooled self-report and quasiobservational data, and on pooled laboratory observational data pre- and posttreatment. Over time it was found that some gains were more durable than others, with the mean Please rate being the least durable, and with maintenance of treatment gains being strongly related to an initial reduction to within nondistressed limits of the ACQ score. Results of the study of single cases indicated that treatment efficacy was related to the degree of pre-existing personality disturbance, to prior learning regarding the appropriateness of expression of feelings, to the availability of alternatives as predicted by social exchange theory and to the diligence with which new skills were practised., The third study was concerned with the prevention of marital distress. The above training programme was extended to a group of 7 newlymarried couples. Treatment gains and maintenance of gains were evaluated by means of repeated measures on the same test battery as used in Study 2. Measurement was made pretreatment and at 0,3,6,9, and 12 months posttreatment. Results were compared to those of a 7-coup1e waiting-list control group. Analysis of pooled self-report and quasi-observational data by means of multiple analysis of covariance produced a significant pre- to posttraining effect (p<.05), while a comparison of experimental and control groups of repeated measures produced a significant long-term training effect (p <.001). No such effects were found when scores on laboratory observational variables were investigated. Results provided only limited support for the classical behavioural model of marital distress, which had been clearly supported in Study 1, and pointed to a surprisingly high incidence of anxiety in a nonclinical population. Results of the three studies taken collectively suggested that the current behavioural model has oversimplified the development of marital distress. An attempt was made to revise the model, taking account of the contribution made by dysfunctional internal stimuli such as anxiety and negative self-statements, and making explicit the relevance of communication concepts to the behavioural model. Progress toward the development of a general theory of interpersonal relationships was explored, and the implications for society of such progress and of behavioural theory in particular were discussed.