About 1300 groundnut fertilizer trials have been conducted throughout Uganda. Little of the variation in final yield and response to P fertilizer in individual trials could be accounted for by soil properties, although crop leaf P levels at flowering were often significantly related to an index of soil phosphate availability. However, average groundnut yields and responses to P fertilizer at centres where up to 30 trials were carried out every season for several years were found to be closely related to the average level of soil organic matter, intensity of previous cultivation and average hours of sunshine received per day. An index of soil phosphate availability was significantly related to all these factors which appear to control the availability of phosphate through their influence on soil organic matter mineralization. Average leaf P level at centres were significantly (negatively) related to rainfall and variation in organic matter mineralization due to weather would explain why the performance of groundnuts in individual trials was poorly related to fixed soil properties. Mean responses to P fertilizer were closely related to mean leaf P levels, which indicates that absence of a response was due to an adequate soil phosphate supply and not to fixation of the applied fertilizer.