Misleading results can arise from using the DIFFUSE computer code to model hydrogen uptake and release by metals. This is due to the fact that it explicitly uses the Baskes model for the hydrogen recombination coefficient. These points are illustrated using recent results on the release of deuterium implanted in beryllium as an example. These data are reinterpreted using the PERI computer code, which is less model dependent than DIFFUSE. The new value for the pumping parameter D/(2kr) (diffusion coefficient over recombination coefficient) is much more in keeping with the values obtained in other experimental investigations.