During the Gibraltar Experiment in 1985-1986, observations of shear, stratification, microstructure, and acoustic backscatter revealed a complex mixing regime at Camarinal Sill in the Strait of Gibraltar. The mixing is forced by strong semidiurnal tidal currents and weaker, but more steady, baroclinic flows. Flows are characterized as outflow or inflow, nearly equivalent to westward or eastward. Rising water at Gibraltar coincides with outflow at all depths. We observed two modes of outflow at the sill. In May 1986 the transition layer separating Atlantic and Mediterranean water was between 0.75 MPa and 1 MPa when we sampled east of the sill near neap tide. The transition turned down just before reaching the sill crest, where the flow became critical. Increasing shear soon destabilized the transition, producing billows which grew to 30-75 m on the west flank of the sill. The overturning generated intense turbulence, with peak dissipation rates, ∊, exceeding 10-2 W kg-1. This turbulence rapidly thickened the transition from ∆z = 30-50 m east of the sill to ∆z = 130-150 m within 1-2 km west of the sill. From the rate of thickening, we estimate the rate of formation of new transitional water during neap tide as (0.8-1.2) × 105 m3 s-1. This outflow mode did not release eastward-propagating internal bores when outflow turned to inflow, but the inflow did appear to arrest the deep outflow. In May, the largest average dissipation rate over and just west of the sill had the same magnitude as estimates of the net energy lost by adjustment of velocity and mass across the sill during outflow. Other averages are smaller than the energy loss, presumably because the limited sampling severely underestimates the average dissipation rate. During outflow, strong shears were observed to rise to 100 times those in the background internal wave field. However, probably owing to spatial offsets between the ship's acoustic Doppler current profiler and our tethered free-fall profiler, the only statistically significant correlations were found between overturning scales measured directly and those estimated from ∊ in the same profile. We observed the other outflow mode in October 1985, when we profiled over the sill near spring tide and the transition was east of the sill, centered near 1.5 MPa. After encountering the sill, the transition rose sharply and remained within 10-20 m of the surface across the sill, plunging steeply downward just past the west edge of the sill. This was accompanied by intense turbulence and at least one rebound of the transition. We did not observe release of eastward propagating bores when inflow turned to outflow, but observations by Armi and Farmer (1988) in April 1986 found bores released at spring tide in regimes similar to the one we observed in October 1985.