Sixty-five percent of the short term variability in southern Mexican precipitation is accounted for by the large-scale circulation. Empirical relationships between sea level and 500 mb circulation fields, and the local precipitation in Chiapas, Mexico, are derived using a neural net. Although much of the rainfall is a result of convective processes, the neural net captures the onset of the precipitation season, and the phase of individual precipitation events. The analysis indicates that both of these aspects of the precipitation regime are controlled to a large extent by the atmospheric circulation. ¿ American Geophysical Union 1992