Because wood density is an important indicator of the end-use mechanical properties of sawn timber, detailed knowledge of its variation is important for optimizing value in processing. The aim of this study was to develop a model for predicting wood density variation in plantation-grown Scots pine (Pinus sylvestris L.) using position in the stem and radial growth increment as independent variables. Seventy trees, ranging in age at 1.3 m from 9 to 99 years, were sampled from 12 even-aged Scots pine stands in Scotland, UK. Pith-to-bark density profiles were obtained on 193 radial samples using scanning X-ray densitometry. We developed a mixed-effects model of average ring density (expressed as basic specific gravity) based on an exponential function of cambial age, height along the stem and annual ring width. Basic specific gravity ranged from 0.274 to 0.697, with a mean of 0.423, and increased rapidly from the pith until approximately rings 20–30, before stabilizing. In addition, there was a weak negative relationship between basic specific gravity and annual ring width, and for a given cambial age, basic specific gravity decreased with sampling height. The fixed effects of the final model were able to explain ∼57 per cent of the within-stem variation in wood density (78 per cent if random effects were included). The final model is intended for integration into a growth, yield and wood quality simulation system.