Wood failure evaluation is the key criterion for predicting the long-term durability of plywood. At present, the conventional visual method for plywood wood failure evaluation is slow and subjective. Evaluations can be influenced by factors such as: room lighting, wood species, sample treatment, and readings from prior samples. An automated wood failure evaluation system using image analysis techniques could potentially be programmed to consider all the variables and respond with consistent wood failure values regardless of the machine operator's experience level. This report describes the results of a six-month study in which a system for automated plywood wood failure determination was compared with conventional visual wood failure evaluation. It was built upon research undertaken in the 1996/97 year in which the feasibility of the approach was initially established. In the research reported previously, a colour optical imaging system was assembled and suitable wood failure algorithms were compiled with promising results. The imaging system was 100 % effective in reproducing sample values. The data were discussed with the project liaisons and a three-month comparison with Canply readings was suggested. In this study, machine evaluation of 4,150 samples was compared with readings of monthly plywood mill quality control samples. The sampling was designed to include all British Columbia plywood mills and all categories of commercial plywood production. The differences in average values for wood failure between human and machine evaluation were found to be less than plus or minus 5% in the majority of cases. In addition, 93 % of ‘set average' readings fell in the plus or minus 10% range of deviation expected of human wood failure readers. Agreement on readings of individual samples within each set was not quite as good with 72% falling in the plus or minus 15% range.