The purpose of this study was to assess how changes in the habit of attending cultural events in the community might predict self-reported health. This study made use of data based on two interviews, 8 years apart, with 3793 adults aged 25-74 years from a simple random sample of the Swedish population. The subjects were interviewed in 1982-1983 and re-interviewed using the same questionnaire in 1990-1991. The setting was a Swedish interview survey of living conditions comprising a random sample of the adult population. Self-reported health status was the main outcome measure. The variables used for control purposes were baseline health status according to the survey of 1982-1983, age (at baseline), type of residence, geographical region of domicile, and socio-economic status (level of education). A correlation was found between perceived poor health and all the independent variables, as well as an influence in the expected direction for all of them; poor education, increasing age and a low degree of urbanization all predicted poorer perceived health. In the full model (including all nine independent variables), those who became culturally less active between the first and second occasion, or those who were culturally inactive on both occasions, ran a 65% excess risk of impaired perceived health compared with those who were culturally active on both occasions. Furthermore, those who changed from being culturally less active to being more active had about the same level of perceived risk as those active on both occasions. These results could be in agreement with a causal influence of stimulation and suggest that cultural stimulation is a 'perishable commodity'. While recruiting new consumers would in that case promote health, continued frequent replenishment of the cultural stimulation may be just as important.