A set of 12 measures based on separate reports by husbands and wives was used to determine whether families of children displaying disturbed behaviour have poorer marital relations than those of children without presenting symptoms. Following an initial discriminant analysis which revealed significant differences between a clinical and non-clinical group of marital pairs, cluster analysis of the total sample yielded three subgroups with greater homogeneity. Two discriminant functions distinguished significantly between these groups; the first function, which accounted for most of the variance, arranged the groups in linear order along a general dimension of marital quality. The second, smaller function separated the middle group from the two extremes which represented mainly clinical and non clinical cases respectively. The quality of marital relations was found to decline from high levels of mutual regard and consensus in the non-clinical group to mutual dissatisfaction in the clinical group and this was greater for wives than husbands. In the middle group, of mixed clinical and non-clinical cases, the wives remained strongly committed to parental and marital roles but dissatisfaction was expressed in the quality of their interpersonal conduct. In general the wife's evaluation of her husband was an important contributor to her level of marital satisfaction and, in turn, to the general quality of the marital relation.