Aid agencies and governments spend more than a billion US$ on entrepreneurship training annually. What have we learned about the effectiveness of training? We review research on entrepreneurship training. Classroom-based training remains the most popular method of training owners and managers of small firms. A meta-analysis shows that the standard training model has modestly positive effects, on average, though the effects imply reasonably high returns on investments in training, given low costs per participant. Innovation on this basic training model has increased in recent years, particularly with regard to content. Both personal initiative and rule-of-thumb training show promise for subsistencelevel enterprises. Individual consulting has shown significant positive effects for larger enterprises, but the model is expensive and markets for consulting do not appear to work well. Selection is important, particularly in matching the type of training with the type of enterprise. There are several seemingly promising approaches to training where definitive evidence is lacking. For example, Kaizen approaches and Incubators and accelerators both appear to have positive effects, though the evidence is limited and, in the case of accelerators, it is unclear as yet whether the effects come primarily from selection or from the content of the programmes themselves.