Managing tropical rain forests is difficult because few long-term field data on forest growth and the impact of harvesting disturbance are available. Growth models may provide a valuable tool for managers of tropical forests, particularly if applied to the extended forest areas of up to 100,000 ha that typically constitute the so-called forest management units (FMUs). We used a stand growth model in a geographic information system (GIS) environment to simulate tropical rain forest growth at the FMU level. We applied the process-based rain forest growth model Formix 3-Q to the 55,000 ha Deramakot Forest Reserve (DFR) in Sabah, Malaysia. The FMU was considered to be composed of single and independent small-scale stands differing in site conditions and forest structure. Field data, which were analyzed with a GIS, comprised a terrestrial forest inventory, site and soil analyses (water, nutrients, slope), the interpretation of aerial photographs of the present vegetation and topographic maps. Different stand types were determined based on a classification of site quality (three classes), slopes (four classes), and present forest structure (four strata). The effects of site quality on tree allometry (height-diameter curve, biomass allometry, leaf area) and growth (increment size) are incorporated into Formix 3-Q. We derived allometric relations and growth factors for different site conditions from the field data. Climax forest structure at the stand level was shown to depend strongly on site conditions. Simulated successional pattern and climax structure were compared with field observations. Based on the current management plan for the DFR, harvesting scenarios were simulated for stands on different sites. The effects of harvesting guidelines on forest structure and the implications for sustainable forest management at Deramakot were analyzed. Based on the stand types and GIS analysis, we also simulated undisturbed regeneration of the logged-over forest in the DFR at the FMU level. The simulations predict slow recovery rates, and regeneration times far exceeding 100 years.