A three-dimensional hydrodynamic model was developed as part of a cooperative U.S. Geological Survey/Massachusetts Water Resources Authority program to study contaminated sediment accumulation and transport in Massachusetts Bay. This report details the development of the model and assesses how well the model represents observed currents and water properties in the bay. It also summarizes circulation and comparative effluent dilution simulations from existing and future Boston sewage outfalls over a three-year period from October 1, 1989 to December 31, 1992. The ECOM-si model, a semi-implicit version of the Blumberg and Mellor (1987) Estuarine, Coastal and Ocean Model, is shown to reproduce many of the important hydrodynamical features of Massachusetts Bay: the seasonal evolution of the pycnocline, the mean flow pattern, and the strength of sub-tidal current fluctuations. Throughout the simulation period, during both vertically well-mixed and stratified conditions, the seasonal statistics of observed currents are well-represented by the model. The model is therefore appropriate for studying the average dilution of sewage effluent and other continuously discharged substances over seasonal time scales. The ability of the model to reproduce individual flow events varies with season and location within the bay. Flow events during unstratified conditions in western Massachusetts Bay are particularly well-represented, indicating that the model is appropriate for studying processes such as the transport of suspended material from the future outfall site due to winter storms. Individual flow events during stratified conditions and in the offshore Stellwagen Bank region, however, are less well-represented due to small length scales (caused by upwelling and river discharge events) coupled with insufficient data to specify open boundary forcing from the Gulf of Maine. Thus while the model might be used to answer issues such as the frequency with which Gulf of Maine river plumes visit the new outfall site, attempting to predict whether a particular plume would visit the outfall site could be problematic. Comparative simulations of effluent discharged from the existing and future Boston outfalls show that the region of relatively high effluent concentrations (1 part effluent to 200 parts sea water) is significantly smaller with the future outfall and is limited to Western Massachusetts Bay during both unstratified and stratified seasons. The region of even higher concentration (1 part effluent to 50 parts sea water) that covers much of Boston Harbor with the existing outfall is non-existent in the future outfall simulation. Additional simulations of chlorination plant failure predict that the offshore location of the future outfall will lead to dramatically lower levels of pathogens at area beaches.