Error scores on the Farnsworth-Munsell 100-hue test were partitioned into those representing red-green and those representing blue-yellow losses. Data from two groups of normal observers were used. One group showed results characteristic of published norms; one group showed superior performance. Both observers showed a correlation between red-green and blue-yellow scores indicative of a strong performance factor in this test. The difference between blue-yellow and red-green scores eliminates their correlated variance and allows evaluation of the axis. Both groups showed an increase in difference scores, with age indicating development of a blue-yellow axis. This increase was significant for the observers characteristic of the norms. We suggest cutoff scores to allow a decision as to whether a given patient shows a blue-yellow or red-green axis.