This paper is part of a broader study of doctors' "sickness certification" practice, which is correlated with, but not the same as, "sickness absence" or "sickness benefits". In order to obtain a total picture, information on sickness certification must be related to the population at risk, i.e. the epidemiology of sickness certification in a total population defined geographically. There is no routine registration system that provides statistics of sickness certification in Norway. Neither is there current information about those of the population who at any one time are entitled to sickness benefits, i.e. the population at risk. The aim of the present paper is to discuss the problems of estimating annual incidences of sickness certification, and to describe the results according to patients' sex, age, and place of residence. The study is based on all "initial certificates" received at the National Insurance Offices in Buskerud county during a period of four weeks in 1985. The population at risk was estimated at 106,019 employed persons aged 16-69 years, and the annual incidence of sickness certification at 580 per 1,000 employed persons per year (females 596, males 568). The highest incidence was found in the age group 20-29 years (females 739, males 741). In the age groups 30-39 and 40-49 years, incidences were significantly higher in females than males. The standardized incidence ratio was significantly lower than average for both females and males in agricultural municipalities, while it was significantly higher than average for females 30-39 years old in urban municipalities. The basis of epidemiological studies of sickness certification used in health services planning and in community medicine is in need of improvement. This challenge is being addressed by the National Insurance Administration in association with the Central Bureau of Statistics in Norway.