The colour of the wood in radiata pine has large variations due to natural factors and also wood processing treatments. Wood colour has been previously studied in many wood-related processing areas, such as pulp and paper, coating, photoyellowing and veneering, while few studies have been done on the colour of fresh, green wood. In this study, the specific light absorption coefficient, k, and the specific light scattering coefficient, s, constitute a colour system that is based on Kubelka-Munk's theory which was used to investigate the inherent wood colour caused by wood properties rather than any other wood processing or environmental factors. An experimental design including five experimental factors (sites, discs, heartwood/sapwood, compression wood/nom1al wood, earlywood/latewood) was designed for investigation of the influence of lignin and wood structural properties on colour. Six wood structural properties (density, specific surface area, cell wall thickness, coarseness, radial and tangential tracheid diameter) were measured using Silviscan analysis. The effects of extractives, moisture, temperature and sample surface texture were measured and kept constant for statistical experiments. Samples were minimized to the microtome level (10 mm x 10 mm x 100 µm) to improve homogeneity. Blue light at 457 run was selected to be the most suitable wavelength to measure brightness, k and s. Due to the small sample mass of microtomes, lignin was analyzed using acetyl bromide instead of the more standard Klason lignin analytical method. The significance of the effect of each wood property on colour parameters was then analyzed in a statistically designed experiment. The results of this study were statistically analyzed and were supportive of a significant relationship (1 % level) between k and lignin, and between s and density. A multiple regression model with two variables was established with the dependent variable being brightness and the independent variables being k and density. The correlation between all variables was analyzed. The significance of variance of all variables in the experimental factors also was analyzed using a SAS package (Section 2.1). A density-b1ightness prediction model was established based on the non-significant variance of k between sites. Further work could be done on a colour-wood property prediction model based on the results in this study. A two-variable model (brightness as a function of specific light absorption and scattering coefficients k and s) can be simplified and applied to predict the colour of radiata pine. This study has shown that the colour of wood can be predicted from knowledge of wood structural properties based on their relationships to k and s as determined. While k can be derived for a wood through its relationship with brightness and s, s itself can be substituted by density, with which it has a very strong correlation. When k is relatively constant, brightness can be substituted by density through s. A second colour criterion, contrast ratio, was defined to quantify the impact of the proportion of latewood on the brightness of wood.