In Part I of this paper (Killworth, 1997), a new eddy parameterization scheme was presented. Here, the scheme is tested by comparing its predictions with those of multi-year averages from an eddy-resolving channel model. Its accuracy is similar to that of a tuned version of previous schemes. However, a tuned version of the new parameterization can reproduce both the long-term average of the eddy-resolving solution as well as the initial slumping of a narrow front. Both tuned schemes reproduced the bolus transport well. The new parameterization reproduces the observed feature that the diffusivity is maximal at mid-depth and minimal at surface and floor.