Data for Pinus radiata D. Don grown in the Australian Capital Territory (ACT) are used to show that annual indices of growth potential can be successfully incorporated into Schumacher projection models of stand basal area growth. Significant reductions in the error mean squares of the models can be obtained by including a simple index such as annual rainfall, but best results were obtained by incorporating estimates of photosynthesis simulated with a detailed process-based model: BIOMASS. In the ACT it was sufficient to estimate the growth index at a single location within the forest estate. Reductions in error mean squares due to the incorporation of temporal variables were about twice as large as those obtained by incorporating spatial variables such as geological substrate, site index or indices of soil development. The gains due to the two classes of variables were approximately additive. The new models improve the descriptive power of the Schumacher model. Short-term predictions made with the models should be more accurate than those obtained with the traditional model and should be particularly useful for updating stand inventories. The new models would be most applicable to regions where there is substantial variation in climatic factors between growing seasons and where the object species is responsive to those factors. A key result is that the temporal variation in the growth indices need not be assessed at each sample plot used to calibrate the model nor each inventory plot to which the model is applied. The temporal variation is regional in nature; consequently, it can be characterised by studies at a relatively few number of sites. This leads the way to new avenues for forest modelling.