Local information on the age‐specific prevalence of disability, distinguishing disability type, is important for planning purposes to inform the provision of specialist services appropriate to the needs of those at different ages. Projections of disability, which require estimates of age‐specific rates of disability as their base, are valuable as they enable planners to prepare for future service demands. However, in the UK, estimates that include detail of disability type, age, and sex are not reliably available for sub‐national areas. This paper evaluates relational models, a technique originally developed for the estimation of age‐specific mortality rates, as a potential solution to this information gap. Relational models are shown to be as successful in capturing sub‐national variability in levels of disability as individual‐level synthetic regression models. The relational approach has important advantages over more commonly used individual‐level synthetic regression models in terms of parsimony of parameters and assumptions. As such, relational models offer a valuable new approach for health researchers and policymakers interested in the estimation of local age‐specific rates of disability and more generally other health‐related characteristics with rates that follow mortality‐like age patterns. Copyright © 2012 John Wiley & Sons, Ltd.