ABSTRACT:The paper addresses the problem of estimating future stand development in heterogeneously structured forests in Sweden; specifically, multi-layered spruce stands and mature pine stands with advanced spruce undergrowth. We first introduce various supporting concepts and models with their empirical databases, model validation and constraints. Secondly, Swedish single-tree growth functions designed for more heterogeneously structured forest are tested using data from inventory plots, a thinning experiment in an uneven-aged forest stand, and yield plots in pristine forest. Future growth of a managed, multi-layered forest was simulated and is compared with other selected functions. Simulation results, expected errors and time constraints are discussed. For most models, projected stand basal area growth deviated 10-20% from the observed growth in individual stands. In single stands, the deviation ranged from 0 to 60%. Validation periods were often 5-15 years, sometimes even more than 30 years. For Swedish single-tree basal area growth functions, on average, a 5% overestimate was found for heterogeneously structured forest across Sweden. Observed growth in a boreal single-tree selection forest was underestimated by 12.5% fifteen years after thinning from above.