Due to high biodiversity in tropical rainforests, tree species are aggregated into functional groups for modelling purposes. In this article the influences of two different classifications of tropical tree species into functional groups on the output of a rainforest model are analysed. The FORMIND model is documented. FORMIND simulates the tree growth of tropical rainforests. The model is individual-based and developed from the FORMIX3 model. In the model, trees compete for light and space in plots of 20 x 20 m in size. A carbon balance is calculated based on the processes of photosynthesis and respiration. Using a tree geometry submodel, typical tree variables (e.g. diameter, height, crown length) are calculated. The mortality process is mainly driven by falling trees and the canopy gaps they create. Trees of the same functional group and diameter class are represented in one cohort. Simulation results for a primary lowland dipterocarp rainforest in Sabah, Malaysia, are discussed. Detailed structural characteristics of the rainforest stands can be analysed, e.g. the simulation results support the hypothesis that rainforests grow in a layer structure. A comparison of results for the aggregation of 436 tree species into five or 22 functional groups respectively, shows that typical results, such as total stem volume or crown closure, achieved with five groups are as accurate as simulations with a huge number of groups. However, some features such as interspecific competition can only be analysed with a huge number (e.g. 22) of functional groups. The model has the potential to analyse extinction processes and spatial structure of gap formations in rainforests. The analysis of logging scenarios could estimate the effects of human impacts in tropical rainforests.