Abstract: Understanding the effects of logging activity on genetic diversity is an important aspect of establishing the sustainability of selective logging management operations in tropical forests. Genetic variation is affected by selective logging directly, through the removal of and damage to trees within the population, and indirectly, through a change in the forest structure and environment in which the remaining population lives. Eco-Gene is a population genetics model applied to tropical forests over a scale of hundreds of hectares. SYMFOR is a modelling framework for individual-based spatially explicit ecological and management models applied to tropical forests over a scale of 0.25 4 ha. We have linked the models to enable simulations using processes involved in both models. To overcome problems of scale, the spatially explicit competition index calculated in SYMFOR simulations has been modelled such that it can be applied at scales representing much larger areas for which the data are not available, as required by Eco-Gene. The competition index is modelled as a distribution on a grid-square basis, and implemented in the linked Eco-Gene/SYMFOR system. Each tree within a grid-square is given a “relative competition” within the distribution, biased according to species. A competition index value is obtained for the tree by transforming the grid-square distribution to be relevant to the size of the tree, and extracting a value according to the tree's relative competition within the distribution. The distribution and each tree's relative competition within it change according to the effects of growth, mortality and logging activity. The model was calibrated using data from the Tapajos region of the Eastern Amazon forest. This paper describes the model, its calibration and validation and the implications of scaling up from an explicit representation to a modelled quantity. (C) 2004 Elsevier B.V. All rights reserved.