Abstract
A metrics-based assessment can predict reasonably well the overall outcome of the Research Assessment Exercise 2008 for social work and social policy and administration in terms of research environment, but not in terms of research outputs. It is not possible to replicate peer review of the research outputs using existing data. It is sometimes argued that citation counts provide an alternative approach that might help research assessment, but it is one fraught with difficulties. Academics did not, in fact, routinely chose to submit their most cited work. At least in this subject, metrics are more suited as handmaiden to peer review than its replacement.