The aim of the present study was to compare the results obtained using two different methods of data collection about caries preventive services provided in general dental practice. A questionnaire was mailed to a random national sample of 479 dentists resident in Norway in January 1985. The sample was divided into two groups by random allocation. All dentists, irrespective of group, were requested to give background information. One group, comprising 287 dentists (GR), was asked to complete a separate form for every adult patient (greater than or equal to 20 yr) treated in the course of 1 day. The demographic characteristics and dental visiting habits of the patients, as well as the number of teeth present, caries lesions and preventive services rendered were recorded. The other group, 192 dentists (GE), was requested to make general estimates of the time spent on caries prevention and the proportion of patients receiving various types of caries preventive services. The dentists were unaware of the methodologic aspect of the survey and everyone received one reminder in order to guarantee anonymity. The estimation method (GE) did not give the expected advantage over the registration method (GR) in response rate (51.7% vs 46.2%, P greater than 0.40), and gave a gross overestimation of the frequency with which adult patients received different types of caries preventive procedures (P less than 0.005). Thus, even though the estimates of the proportion of total treatment time spent on caries prevention were comparable for the two methods, and the estimation approach is labor-saving, it cannot be recommended for the collection of data on caries prevention in the dental office.