There already exists a body of literature on the conduct of classical epidemiological studies. However, guidelines tend towards detailing how such studies should be conducted and may not prepare the reader for the less-than-perfect scenarios that will inevitably be encountered. The Dental Health Services Research Unit in Dundee has been involved in longitudinal studies of dental treatment and dental health since its inception in 1979. The problems encountered in this research are considered under the headings of mounting the studies, samples, data collection, external changes, internal changes, dissemination and curtailment. It is hoped that a description of the often unpredictable problems associated with a particular set of studies will provide an insight which may assist others embarking on analogous projects in health services research.