Increasing emphasis on ecological objectives such as biodiversity in forestry has precipitated interest in multiaged silvicultural methods that create and maintain more complex stand structures than simpler even-aged methods. Foresters often use the Forest Vegetation Simulator (FVS)
growth model to compare competing silvicultural alternatives, which can include multiaged structures. For models such as FVS to be useful for this task, they must realistically simulate structural dynamics in multiaged stands. We evaluated FVS performance for trees in even-aged and multiaged
stands of ponderosa pine (Pinus ponderosa C. Lawson var. scopulorum Engelm.) in the Black Hills, USA, by comparing FVS predictions of dbh increment to measurements from increment cores. Dbh change is the primary element of tree growth estimated by the regional FVS variant used in the Black
Hills and is the basis for all other aspects of tree growth. We found that the dbh growth model used by FVS was more biased and less accurate for trees in multiaged stands than for trees in even-aged stands and that residual error varied systematically with tree size class in multiaged stands,
which could affect stand dynamics simulation. Performance improved for trees in multiaged stands when predictions were adjusted with a model that included crown ratio (CR), which is related to photosynthetic capacity and tree growth. A CR-based adjustment did not improve model predictions
for trees in even-aged stands, presumably because CR was less variable than in multiaged stands and because stand structures were more similar to those used for model development. This study illustrates the potential for empirical forest growth models such as FVS to behave in unexpected ways
when they are used for stands and trees that are dissimilar from those used for model development. It also highlights the importance of continued model testing because growth models are increasingly used to develop silvicultural alternatives for structurally complex stands.