Seedlings of five tropical trees, Cecropia obtusifolia, Myriocarpa longipes, Piper auritum, Senna multijuga and Trichospermum mexicanum, were grown both as individuals, and in competition with each other at ambient (350) and two levels of elevated CO2 (525 and 700 μl l-1) for a period of 111 days. Growth, allocation, canopy architecture, mid-day leaf water potential and soil moisture content were assessed three times over this period for individually grown plants, and at the end of the experiment for competitively grown plants. In addition, leaf photosynthesis and conductance were assessed for the individually grown plants midway through the experiment, and light profile curves were determined for the competitive arrays at three stages of development. Elevated CO2 did not affect photosynthesis or overall growth of the individually-grown plants but did affect canopy architecture; mean canopy height increased with CO2 in Piper and Trichospermum and decreased in Senna. Stomatal conductance decreased slightly as CO2 increased from 350 to 525 μl l-1 but this had no significant effect upon whole plant water use of leaf water potential. Soil moisture content for the individuals increased marginally as CO2 increased, but this did not occur in the competitive arrays. There was a marked effect of CO2 upon species composition of the competitive arrays; Senna decreased in importance as CO2 increased while Cecropia, Trichospermum and Piper increased in importance. Stepwise regression analysis using competitive performance as the independent variable, and the various morphological and physiological parameters measured on the individually grown plants as independent variables, suggested that canopy height was the single most important variable determining competitive ability. Also significant were photosynthetic rate (particularly at low light levels) and allocation to roots early in the experiment. Light profiles in the canopy revealed that less than 15% of incident light penetrated to the level of mean canopy height. Results suggest that competition for light was the major factor determining community composition, and that CO2 affected competitive outcome through its affect upon canopy architecture.