Most of the existing emission models developed from small-scale chamber tests assume complete mixing in the chamber throughout the test period. This paper examined this assumption using a Computational Fluid Dynamics (CPD) model. The model simulated the three-dimensional air velocity profiles and Volatile Organic Compound (VOC) concentration distributions from wood stain in a well-designed mixing chamber of 1.0X0.8X0.5 m3. The model used measured data to determine the time-dependent voc surface concentrations of wood stain. The CFD results show that the VOC concentrations in the test chamber were not uniform in the early stage (about 18 minutes). The first-order decay model using the complete mixing assumption will underestimate the Total VOC (TVOC) emission rates by 65% and 59% in the first 3 minutes and next 15 minutes, respectively. Since wood stain emitted about one third of the VOCs in the first 18 minutes, the impact of incomplete mixing in the early period is significant for calculating the material emissions. Furthermore, the mass transfer coefficient of TVOC calculated by CPD is also compared with that calculated by analogue theory and that calculated by experimental correlation.