A basal area growth model was developed to predict the growth of individual trees in second-growth black spruce (Picea mariana (Mill.) BSP) stands on northeastern Ontario peatlands. The data were derived from stem analysis trees collected in 1985 and 1986 from stands harvested 47-68 years earlier. For a period starting from the date of data collection and going back to 10 years from the harvesting, tree basal area growth, diameters, and stand characteristics were retrospectively calculated at 5-year intervals. To estimate previous mortality, self-thinning relationships for black spruce were applied. In the model, 5-year basal area growth of a tree was expressed as a function of tree diameter, stand-level competition, tree-level competition, and peat thickness. There was considerable change in the growth-size relationship over time. A random parameter approach was applied in model construction to account for the spatial and temporal correlations of the observations. The proposed model explicitly incorporates factors normally included in a "random error" term and, therefore, should provide more sensitive tests of the contributions of the various factors to growth prediction. The estimated model showed only slight bias against the modeling data and the predicted stand basal area development was comparable with that given in other studies.