BACKGROUND: The use of clinical guidelines in general practice is often limited. Research on barriers to guideline adherence usually focuses on attitudinal factors. Factors linked to the guideline itself are much less studied. AIM: To identify characteristics of effective clinical guidelines for general practice, and to explore whether these differ between therapeutic and diagnostic recommendations. DESIGN OF STUDY: Analysis of performance data from an audit study of 200 general practitioners (GPs) in The Netherlands conducted in 1997. SETTING: Panel of 12 GPs in The Netherlands who were familiar with guideline methodology. METHOD: A set of 12 attributes, including six potential facilitators and six potential barriers to guideline use, was formulated. The panel assessed the presence of these attributes in 96 guideline recommendations formulated by the Dutch College of General Practitioners. The attributes of recommendations with high compliance rates (70% to 100%) were compared with those with low compliance rates (0% to 60%). RESULTS: Recommendations with high compliance rates were to a lesser extent those requiring new skills (7% compared with 22% in recommendations with low compliance rates), were less often part of a complex decision tree (12% versus 25%), were more compatible with existing norms and values in practice (87% versus 76%), and more often supported with evidence (47% versus 31%). For diagnostic recommendations, the ease of applying them and the potential (negative) reactions of patients were more relevant than for therapeutic recommendations. CONCLUSION: To bridge the gap between research and practice, the evidence as well as the applicability should be considered when formulating recommendations. If the recommendations are not compatible with existing norms and values, not easy to follow or require new knowledge and skills, appropriate implementation strategies should be designed to ensure change in daily practice.