This paper describes the development of models to predict wood density in Sitka spruce (Picea sitchensis (Bong.) Carr.) growing in Britain. The models were designed to be practical and widely applicable and useful for large-scale resource mapping and investigating the impact of silvicultural treatments. Two models (and a variant of the second) were developed using data obtained from a 52-year-old spacing trial in Clocaenog Forest, North Wales, and a 32-year-old re-spacing experiment in Kershope Forest, Northern England. The model equations were based on models used to describe the within-tree variation in the wood density of Norway spruce (Picea abies L. Karst.) in France and are functions of ring width and ring number from the pith. The models were able to explain up to 48 per cent of the variability in density from Clocaenog but only 27 per cent of the variability in the density from Kershope, although they were able to accurately represent the radial trends in wood density. Some of the difficulties in explaining the variation in density may be due to the differences in the site conditions, silviculture and age of the trees at the two sites, but it is clear there is significant inter-tree variation in the data.