During the summer of 1992, isoprene emissions were measured in a mixed deciduous forest near Oak Ridge, Tennessee. Measurements were aimed at the experimental scale-up of emissions from the leaf level to the forest canopy to the mixed layer. Results from the scale-up study are compared to different canopy models for determining the leaf microclimate as input to isoprene emission algorithms. These include (1) no canopy effects, (2) a simple vertical scaling canopy model with a leaf energy balance, and (3) a numerical canopy model which accounts for leaf-sun geometries, photosynthesis, respiration, transpiration, and gas transport in the canopy. Initial evaluation of the models was based upon a standard emission rate factor of 90 μgC g⁻¹ hr⁻¹ (0.42 nmol g⁻¹ s⁻¹) taken from leaf cuvette measurements and a biomass density factor of 203 g m⁻² taken from biomass surveys and a flux footprint analysis. The results indicated that predicted fluxes were consistent among the models to within approximately ±20%, but that the models overestimated the mean flux by about a factor of 2 and overestimated the maximum observed flux by 30 to 50%. Adjusting the standard emission factor and biomass density each downward by 20% yielded predicted means approximately 20% greater than the observed means and predicted maxima approximately 25% less than the observed maxima. Accounting for changes in biomass density as a function of direction upwind of the tower improved the overall model performance.