The chance that a couple who have tried to conceive for 12 months will succeed without assisted conception treatment is still higher than the chance that the same couple will benefit from treatment. In this context, it is important to assess the chance that a treatment-independent or 'spontaneous' pregnancy will occur in a couple whose wish for a child is unfulfilled. Prognostic models can be useful in this assessment. In recent years, prognostic models have been published both for the occurrence of 'spontaneous' pregnancy and for pregnancy after in vitro fertilization. This article discusses the theoretical aspects of prognostic modelling and assesses whether the current prognostic models are good enough to justify their use in clinical practice. The performance of existing models for the prediction of spontaneous conception was found to be acceptable on internal as well as on external validation. However, the performance of the existing models predicting IVF outcome was found to be disappointing on the few occasions on which such external validation has been performed.