OBJECTIVE To study the association between informant stress and appraisal of patients' cognitive functioning as reported by the Informant Questionnaire on Cognitive Decline in the Elderly--IQCODE. METHODS Routinely collected data from a geriatric outpatient department (207 dyads) during the years 1995-1998 were analysed. Relative stress scale (RSS) has been categorised for possible low, intermediate and high risk of psychiatric morbidity and caregivers were combined to four groups (female and male spouses and female and male non-spouses, respectively). The relationship between IQCODE (dependent) and categorised RSS and informant groups and patient age was further studied by means of the general linear model (GLM-UNIANOVA). RESULTS In general, spouses reported better cognitive functioning than non-spouses. There was a significant association between IQCODE and RSS (p < 0.001), and the composite variable informant group and informant gender (p < 0.001). The main effect of the interaction term RSS x informant group + informant gender was not significant. Post hoc test, however, revealed a significant effect of the interaction term RSS x female spouses (p < 0.001) on IQCODE. CONCLUSIONS IQCODE is associated with informant stress. Categorisation of RSS score into groups of low, intermediate and high risk for psychiatric morbidity can be a valuable contribution to a more meaningful application of RSS in general practice.