The substantive aim of this research was to identify the factors which affected GPs’ use of the decision support software PRODIGY (Prescribing RatiOnally with Decision support In General practice studY) during the second trial phase of its evaluation. The methodological aim of the paper is to show how problems of multi-collinearity, self-selection and unobserved factors, which typically arise in observational studies, were overcome via a combination of study design and application of the correct statistical procedures. The exercise finds that GP and practice-specific factors, as well as guideline factors, have a significant role in determining the individual GP’s use of PRODIGY.