To describe the basis on which our knowledge of frequent attendance in general practice rests and to propose recommendations for further research on frequent attenders (FAs). The literature review (finished February 2004) encompassed peer-reviewed articles in English describing contacts with general practice in terms of frequency. Searches were performed in the Medline, CINAHL, EMBASE, PsycINFO, Social Sciences Expanded Index and ISI Citation databases with additional searches in reference lists and the 'related articles' function in the ISI Citation database and Medline. General practice. Sixty-one articles (54 studies). The articles were assessed according to the following design variables: setting; definition of FAs; sampling; sample size; control groups; study aim; study design; data sources; effect measure; and main results. There was no generally accepted definition of frequent attendance. Research designs differed substantially. Eight articles gave sufficient information on all design variables. The top 10% of attenders accounted for 30-50% of all contacts, and up to 40% of FAs were still FAs the following year. More than 50% of FAs had a physical disease, more than 50% of FAs suffered from psychological distress, social factors (low social support, unemployment, divorce) were associated with frequent attendance in more than 50% of FAs, multiproblems (physical, psychological and social) were found in one-third of FAs, and frequent attendance was associated with increasing age and female gender. The diversity of designs, definitions and methods in the current literature on FAs in general practice hampers comparison of their precision, validity and generalizability, and calls for cautious interpretation and adoption of a common, generally acceptable definition in future studies.