Fast pyrolysis of woody biomass is a promising process capable of producing renewable transportation fuels to replace gasoline, diesel, and chemicals currently derived from nonrenewable sources. However, biomass pyrolysis is not yet economically viable and requires significant optimization before it can contribute to the existing oil-based transportation system. One method of optimization uses detailed kinetic models for predicting the products of biomass fast pyrolysis, which serve as the basis for the design of pyrolysis reactors capable of producing the highest value products. The goal of this work is to improve upon current pyrolysis models, usually derived from experiments with low heating rates and temperatures, by developing models that account for both transport and pyrolysis decomposition kinetics at high heating rates and high temperatures (>400 °C). A new experimental technique is proposed herein: spatiotemporally resolved diffuse reflectance in situ spectroscopy of particles (STR-DRiSP), which is capable of measuring biomass composition during fast pyrolysis with high spatial (10 μm) and temporal (1 ms) resolution. Compositional data were compared with a comprehensive 2D single-particle model, which incorporated a multistep, semiglobal reaction mechanism, prescribed particle shrinkage, and thermophysical properties that varied with temperature, composition, and orientation. The STR-DRiSP technique can be used to determine the transport-limited kinetic parameters of biomass decomposition for a wide variety of biomass feedstocks.