Field data taken in the Gulf of Cadiz have been analyzed to describe some aspects of the momentum and energy balance of the Mediterranean outflow. A crucial component of the momentum balance is the total stress (entrainment stress and bottom drag), which has been estimated from a form of the Bernoulli function evaluated from density and current observations. For the first 60 km west of the Camarinal Sill the outflow was confined within a narrow channel on the continental shelf. At about 70 km downstream the outflow crossed over the shelf-slope break and began to descend the continental slope. The buoyancy force increased substantially, and the outflow underwent a geostrophic adjustment, albeit one heavily influenced by mixing and dissipation. The current direction turned 90 degrees to the right at a near-inertial rate. In this region, the estimated geostrophic velocity greatly underestimated the actual current, and the estimated curvature Rossby number was about 0.5. Current speeds were in excess of 1 m s -1 and the total stress was as large as 5 Pa. The entrainment stress, estimated independently from property fluxes, reached a maximum of about 1 Pa, or considerably smaller than the inferred bottom stress. By about 130 km downstream, the current was aligned approximately along the local topography. The current amplitude and the estimated stress were then much less, about 0.3 m s -1 and 0.3 Pa. The entrainment stress was also very small in this region well downstream of the strait. This slightly damped geostrophic flow continued on to Cape St. Vincent where the outflow began to separate from the bottom. Bottom stress thus appears to be a crucial element in the dynamics of the Mediterranean outflow, allowing or causing the outflow to descend more than a kilometer into the North Atlantic. In the regions of strongest bottom stress the inferred drag coefficient was about 2 - 12 (× 10 -3) depending upon which outflow speed is used in the usual quadratic form. Entrainment stress was small by comparison to the bottom stress, but the entrainment effect upon the density anomaly was crucial in eroding the density anomaly of the outflow. The observed entrainment rate appears to follow, roughly, a critical internal Froude number function.