A laterally intergrated, time-dependent, numerical model of Burrad Inlet and Indian Arm is compared to time series of cyclesonde velocity and scalar data that have 10 m vertical resolution. As long as deepwater renewal is not occurring, the model can reproduce the observed M2 and K1 tidal response in the basin of Indian Arm. Also, although the rate of freshwater inflow into Indian Arm and Burrard Inlet is not large, the simulated tidal response is noticeably improved if freshwater inflow is taken into account. When deepwater renewal is observed to occur, the model also simulates renewal, but it does so poorly, and the ability of the model to accurately simulate the M2 and K1 tide degrades significantly. ¿American Geophysical Union 1991