The Indonesian Throughflow is analysed in two runs of the OCCAM 1/4 degree global ocean model, one using monthly climatological winds and one using ECMWF analysed six-hourly winds for the period 1993 to 1998. The long-term model throughflow agrees with observations and the value predicted by Godfrey's Island Rule. The Island Rule has some skill in predicting the annual signal each year but is poor at predicting year to year and shorter term variations in the total flow, especially in El Niño years. <br><br> The spectra of transports in individual passages show significant differences between those connecting the region to the Pacific Ocean and those connecting with the Indian Ocean. On investigation we found that changes in the northern transports were strongly correlated with changes in the position of currents in the Celebes Sea and off Halmahera. Vertical profiles of transport are in reasonable agreement with observations but the model overestimates the near surface transport through the Lombok Strait and the dense overflow from the Pacific through the Lifamatola Strait into the deep Banda Sea. In both cases the crude representation of the passages by the model appears responsible. <br><br> In the north the model shows, as expected, that the largest transport is via the Makassar Strait. However this is less than expected and instead there is significant flow via the Halmahera Sea. If Godfrey's Island Rule is correct and the throughflow is forced by the northward flow between Australia and South America, then the Halmahers Sea route should be important. It is the most southerly route around New Guinea to the Indian Ocean and there is no apparent reason why the flow should go further north in order to pass through the Makassar Strait. The model result thus raises the question of why in reality the Makassar Strait route appears to dominate the throughflow.