Crown traits and competition attributes have an important effect on tree radial increment. Relationships among these elements are modeled using the distributions of the crown characteristics in a given calendar year, but these patterns can differ over time. The suitability of the patterns during recovery and normal growth was investigated using silver fir, Abies alba Mill., in old-growth forests. Generalized additive models (GAMs) for silver fir in the older (OG, trees aged 136–300 years) and younger (YG, trees aged 45–135 years) generations were developed. To test the validity of these GAMs, field data sets representing silver fir recovery and normal growth were used. For silver fir in OG, crown transparency had the largest effect on tree growth, explaining more than 25% of the variance. For silver fir in YG, relative crown length had the largest effect on tree growth, explaining more than 15% of the variance. The absolute relative prediction errors, AREmin and AREmax, were less than 0.03 and 1.50 mm, respectively. The developed GAMs are suitable during recovery and normal growth, but the GAMs were fitted to a relatively small area, neglecting climatic gradients and different disturbance types. This type of investigation should be continued on a larger scale.