OBJECTIVE To describe health service consumption and to find out whether a series of possible explanatory variables can help us to predict the number of contacts among children. METHODS A prospective study following a cohort of children during the first four years of life. METHODS A municipality in southern Norway. METHODS 183 children born in the community from October 1979 to and including December 1980. METHODS Direct and indirect encounters with general practitioners (GPs) from the bills to the national insurance office, outpatient encounters and hospital admissions from the hospital files, and episodes of illness as reported in postal questionnaires to the parents. RESULTS Each child had on average 10.6 GP contacts (2.6 per year), and the frequency of contacts decreased as they grew. Telephone contacts were almost as frequent as consultations at the doctor's surgery, and the doctors made more home visits to the youngest children. Of the contacts, 4/5 took place in the day time, but most of the home visits took place during the evening/night/weekend. Each child had 1.9 hospital contacts in 4 years, i.e. 0.3 admissions and 1.6 outpatient department contacts. The parents reported an average of 11.5 episodes of illness during the period, more than twice as many in the first year as in the last year. Information about sex, duration of breast feeding, smoking in the family, family history of allergy, parents' education, whether in kindergarten, and presence of psychosocial problems cannot help in predicting health care consumption. CONCLUSIONS The low number of contacts with GPs can partly be explained by the fact that the children also have contact with a well-baby clinic. In our study, a series of possible explanatory factors did not help us to predict the amount of health care consumption during the first four years of life.