The mean and seasonal variations in transport through and within the Southeast Asian seas are investigated using a series of simple models. The results are compared with results from a fine-resolution, 3D, numerical simulation of the global circulation from 1987 to 1995 [Parallel Ocean Climate Model (POCM)]. For the mean circulation, the models are based on Sverdrup dynamics with the circulation around each island calculated according to the island rule with overlapping islands taken into account. Assuming all of the passages are wide and deep yields an archipelago circulation vastly at odds with observations. A large westward transport through Torres Strait provides the throughflow between the Pacific and Indian Oceans. A large westward transport through Luzon Strait passes southward through the South China Sea into the Sulu Sea and exits into the Pacific Ocean through the Celebes Sea. There is a northwestward transport through the remainder of the archipelago. By successively blocking straits under the assumption that frictional effects are sufficient to arrest flow in the strait, an understanding is built up of why the mean circulation in the archipelago is as observed and as simulated in POCM. For example, blocking Torres Strait yields a more realistic circulation with southward flow in the archipelago. Greater realism is achieved by blocking off the South China Sea, so making the dominant pathway for the throughflow from the Pacific westward through the Celebes Sea and southward through Makassar Strait. The weak throughflow in POCM (7.5 × 106 m3 s−1) is found due to the wind stresses derived from the European Centre for Medium-Range Weather Forecasts 10-m twice-daily winds, which are much weaker than the Hellerman and Rosenstein climatology (HR) used in previous studies. Also, POCM’s throughflow is wholly fed by the South Equatorial Current rather than predominantly by the Mindanao Current, as found in models forced by HR climatology. Analysis of the wind stress datasets and that of the Florida State University from 1961 to 1995 shows that the latitude of the zero-Sverdrup-transport streamline near the Pacific entrance to the Celebes Sea has shifted poleward over the decades, so decreasing the absolute amount originating from the Mindanao Current. Regarding the seasonal cycle, there is negligible transport below 500 m at annual period within the archipelago in POCM, which suggests that the numerous islands and sills within the archipelago enhance the adjustment to the applied wind stress locally. Assuming a local Sverdrup balance, island-rule-based models of the archipelago show that forcing by wind stresses over the archipelago and Australia give reasonable agreement with POCM for the amplitude of the annual harmonic in depth-integrated transport. Better agreement in phase within the straits and seas is obtained by recognizing that frictional effects within certain straits enables the influence of wind stress variations to be felt in directions other than just to the west, as in the original island rule. It is further noted that the adjustment to semiannual period wind-stress forcing is incomplete within the seas;there is no local quasi-equilibrium response. In POCM, the archipelago fills above 500 m in February–June and September–November, and drains in the remaining months. There is compensating flow below. Also, seasonal variability of the currents in the west Pacific is not sufficient to alter significantly the gyre closure in the west Pacific, and the depth-integrated throughflow is fed by the South Equatorial Current throughout the year, either via a western boundary current or a broad zonal jet.