Forest growth simulators go beyond a mere tabulation of empirical measurements by employing biometric models that functionally describe the dependence of forest growth of the initial forest structure, growth conditions and management regime. This makes them very flexible and allows predicting growth reactions for unknown and/or complex forest growth scenarios. When simulation outcomes are to be used in silvicultural strategic planning, the results are of direct and delicate importance, and the correct simulator performance must be ascertained. This is especially so when the considered forest situation differs from the forest data used to parameterise the model (e.g. different geographical region). In this article, the forest growth simulator SILVA (version 2.2) was validated for 55 long-term experimental plots of mature mixed Silver fir–Norway spruce stands in southwest Germany ( Picea abies, Abies alba). The evaluation was restricted to the upper canopy trees during the survey period 1989–2004. Following the general evaluation criteria for ecological models from [Vanclay, J.K., Skovsgaard, J.P., 1997. Evaluating forest growth models. Ecol. Mod. 98, 1–12], a specific methodology was developed to evaluate the simulated height and diameter growth on the basis of forest growth principles. The qualitative analysis proved the SILVA growth algorithms to be in accordance with physiologically based standard growth equations. The quantitative evaluation was limited by incomplete knowledge of the site conditions. To overcome this problem, the experimental plots were regarded as a “heterogeneous growth series” which allows analysing the growth behaviour in a more general way. It could be shown that for the given data set, the SILVA simulations produced an overestimation of height growth (median: +61% spruce, +12% fir), and an underestimation of diameter growth and competition sensitivity (median: −16% spruce, −70% fir). The errors partially compensated in the volume growth resulting in an overall over-/underestimation of +9% for spruce and −58% for fir (median). The unbalanced height and diameter growth cannot be compensated by a change in the site conditions because this affects both height and diameter growth either positive or negative. Hence, an adjustment of selected parameterisation values appears to offer the best solution to adapt SILVA to the considered forest situation. This approach of adaptive parameterisation is discussed against a more general background of deductive vs. inductive forest growth modelling.