A comparison is made between a big-leaf model (i.e., without details of the canopy profile) and two multilayer models (i.e., with details of the canopy profile) to estimate daily canopy photosynthesis of a sugar maple (Acer saccharum Marsh.) stand. The first multilayer model uses the distribution of leaf area by leaf mass per unit area (LMA) classes, the observed relationships between the parameters of a photosynthesis-irradiance curve and LMA, and the relationship between relative irradiance and LMA to estimate canopy photosynthesis. When compared with this model, the big-leaf model underestimates daily canopy photosynthesis by 26% because of an assumed proportionality between photosynthetic capacity and relative irradiance, a proportionality that is inconsistent with our data. The bias induced by this assumption is reduced when the big-leaf model is compared with the second multilayer model which, in addition to the assumptions made for the first multilayer model, accounts for the sunlit and shaded fractions of leaf area. The residual bias is almost eliminated when the big-leaf model is run using a weekly averaged irradiance. It is likely, however, that this is the result of a compensating bias that, in this particular case, compensates for the initial bias introduced by the proportionality assumption. It is also shown that canopy photosynthesis can be represented by spatially inexplicit multilayer models that use leaf mass per area as a covariable to describe leaf characteristics and environment. Such models represent an interesting alternative to the biased big-leaf approach.