A tracer release experiment to study diapycnal and isopycnal mixing was performed in Santa Cruz Basin in 1988--1989. Approximately 1450 g of fulfur hexafluoride were injected close to an isopycnal surface at 1500 m depth in May 1988. This depth is midway between the bottom of the basin at 1920 m and the sill at 1084 m. The diapycnal dispersion measured during the first few months, while the tracer was still mixing to the walls of the basin, yielded a diapycnal eddy diffusivity of 1.0 cm2 s-1, within a factor of 1.5. This is approximately 4 times larger than the diffusivity found by Ledwell and Watson (1991) in the interior of Santa Monica Basin, where the buoyancy frequency was 5 times greater. The tracer was well mixed to the walls 6 months after the injection, and at that time the diapycnal spread of the tracer was greater near the walls than in the interior. The rate of diapycnal dispersion of the tracer increased dramatically after the tracer had mixed to the walls of the basin, indicating enhanced boundary mixing. An effective basin-wide diapycnal diffusivity of 10 cm2 s-1 would be required to support the fluxes of tracer observed. A one-dimensional model for tracer and heat diffusion does not simulate the results very well, however, probably because of the finite mixing time between the boundary region and interior and because of the complexity of the processes in the boundary region. ¿ American Geophysical Union 1995.