What sort of climate information do farmers need? And how can we provide them with forecasts that will be truly useful? This is a complex question, involving a range of fields and requiring various types of data in order to make forecasts on the time and spatial scales on which farmers work and adjust them to the agricultural and technical contexts in which they are to be used. A comparative study conducted in Cameroon, Kenya and Argentina by scientists from an interdisciplinary research group, of which CIRAD is a member, recently showed that the climate data that are really useful to farmers vary substantially depending on the region and on farming practices.