Indonesia is impacted by severe droughts that cause major food shortages over much of the country, and have long been linked with El Niño events in the tropical Pacific Ocean. Despite seasonal forecasts based on relatively complex climate models, the 1997-1998 'El Niño of the century' was still followed by huge shortfalls in crop production (e.g. a loss of 3 million tons of rice in Java alone), reflecting the large gap that can exist between these forecasts and their actual utilization in agricultural planning. Alternatively, simple predictive models of rainfall and crop yields for Indonesia and other Indian Ocean rim countries have utilized an index of equatorial Pacific sea surface temperatures (Niño-3.4 SST), and related this index directly to indices of rainfall and crop productivity. However, these latter models have not included climatic information from the Indian Ocean, also implicated as a likely cause of drought in western Indonesia. Here we show that an index of Indian Ocean SSTs, when combined with the previously utilized Niño-3.4 SSTs, provides a significantly more accurate model of Sept-Dec drought (PDSI) over Java, Indonesia, than Niño-3.4 SSTs alone (ar 2 64.5% vs 37.9%). Based solely on data for the month of August, this model provides the best tradeoff between model skill and adequate lead time for the Sept-Dec rice planting season. Such simple models can be used to generate readily usable early warning forecasts of drought and crop failure risk in Indonesia, particularly in the western part of the country that is most influenced by Indian Ocean climate variability.