We analyzed the probability density function (PDF) of velocity and temperature differences in the canopy sublayer of Amazonia based on Tsallis' generalized thermostatistics theory. We show that such a theory provides an accurate framework for modeling the statistical behavior of the inertial subrange above and below the canopy. For this, we compared the experimental PDFs with the theoretically predicted ones. The data were measured during the wet season of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA), which was carried out during the months of January--March 1999 in the southwestern part of Amazonia region. Measurements were made simultaneously at different heights in a 60 m micrometeorological tower located in the Biological Reserve of Jaru (10¿04'S, 61¿56'W), Brazil. The fast response wind speed measurements, sampled at 60 Hz rate, were made using three-dimensional sonic anemometers at the heights of 66 m (above the canopy) and 21 m (below the canopy). The results showed good agreement between experimental data measured above the canopy forest and Tsallis' generalized thermostatistics theory. For below canopy data, the agreement between experimental and theoretical PDFs was fairly good, but some distortion was observed. This is probably due to some peculiar characteristics of turbulent momentum transfer process inside the forest crown. Discussion is presented to explain these results. Conclusions regarding the absence of universal scaling in the inertial subrange are also presented in the context of the entropic parameter of Tsallis' theory.