The aim was to study whether an audit of treatment of infections in general practice resulted in changed prescribing habits. In 1995-1996 forty-six general practitioners (GP's) from the County of Roskilde participated in an audit regarding infectious diseases (incl. course participation and preparation of treatment guidelines). The effect evaluation was done on the basis of 1) two self-registrations of antibiotic prescriptions carried out with one year's interval, and 2) prescribing data from the National Insurance database collected over two periods, before the first and second self-registration respectively. The number of patients not receiving antibiotics increased significantly from 47.2% to 52.4% after intervention. The self-registration did not show any change in choice of antibiotics, while the registry data showed a shift from broad-spectrum to narrow-spectrum penicillin. This change was, however, also found among the GP's, who did not participate in the audit. The study demonstrated that audit can result in changes in prescribing patterns, but at the same time emphasizes the need for inclusion of external data sources and control groups in the evaluation of intervention effects.