Summary Data from the permanent plots in natural regrowth forests of mountain ash (Eucalyptus regnans F. Muell.) and alpine ash (E. delegatensis R.T. Bak.) in Victoria were used to evaluate the accuracy of projection by the growth model, STANDSIM. In general, where the stand ages used to initiate STANDSIM's projection were 15 yr or more, the errors (biases) for one-year projections would be expected to be within ±1% of the observed values for the projected future stand densities, ±0.5% of the observed values for the projected stand basal areas, and within ±1% of the observed values for the projected total product volumes. These resultant error estimates indicated that the accuracy of the future yields projected from STANDSIM model was generally in an acceptable range, if the projection intervals were less than 50 yr and projections were initiated at a stand age of 15 yr or older. However, the projected yields from this model would have a lower precision if the projections were initiated at stand ages of less than 15 yr. As indicated by a previous researcher, the small-diameter trees were under-represented in the projected diameter distributions. This problem may indicate that the STANDSIM model overestimates the competition-induced mortality of natural ash eucalypt forests. However, our investigation did not indicate that this would result in a direct underestimation of pulpwood and overestimation of sawlog yield.