- BACKGROUND: Several models have been published for the prediction of spontaneous pregnancy among subfertile patients. The aim of this study was to broaden the empirical basis for these predictions by making a synthesis of three previously published models. METHODS: We used the original data from the studies of Eimers et al. (1994), Collins et al. (1995) and Snick et al. (1997) on couples consulting for various forms of subfertility. We developed a so-called three-sample synthesis model for predicting spontaneous conception leading to live birth within 1 year after intake based on the three data sets. The predictors used are duration of subfertility, women's age, primary or secondary infertility, percentage of motile sperm, and whether the couple was referred by a general practitioner or by a gynaecologist (referral status). The performance of this model was assessed according to a 'jack-knife' analysis. Because the post-coital test (PCT) was not assessed in one of the samples, a synthesis model including the PCT was based on two samples only. RESULTS: The ability of the synthesis models to distinguish between women who became pregnant and those who did not was comparable to the ability of the one-sample models when applied in the other samples. The reliability of the predictions by the three-sample synthesis model was somewhat better. Predictions improved considerably by including the PCT. CONCLUSIONS: The synthesis models performed better and had a broader empirical basis than the original models. They are therefore better suitable for application in other centres.