Factors affecting the supply of air to leaves enclosed in assimilation chambers have been studied using models and leaves of apple and plum. The rate of CO2 assimilation of a leaf does not increase linearly with increasing rate of air flow. Explanations of this and other observations are put forward in terms of the boundary-layer concept and diffusion paths. Comparisons are made with free-air conditions, and it is concluded that air should be supplied at the same rate (ml/min) for leaves of different size in the same chamber. Three cup designs were also studied. A method is described for making valid comparisons of assimilation rates per unit area by extrapolation to an infinitely high rate of air flow. Measurements of different sized leaves with chambers and cups can be compared in this way. Rates expressed per unit area can be correctly compared only if the resistance to diffusion outside the leaf is negligible, or under special conditions. Assimilation rates measured with rates of air flow giving CO2 availability equivalent to that occurring in static free air should be expressed per unit perimeter. Methods of correction of assimilation rates for the depletion of carbon dioxide from the air stream are considered.