OBJECTIVE To compare blue-on-blue differential contrast perimetry (BB), in accordance with E. Land "Retinex" theory, with white-on-white (WW) and blue-on-yellow (BY) perimetries on normal subjects. METHODS An Octopus 101 perimeter was modified for BB perimetry, using a 4cd/m2 background and stimulus Goldmann size V. Fifty healthy subjects (10 per decade, from 20 to 70 years) were examined twice with each type of perimetry (WW, BB, BY) using the TOP strategy. RESULTS The results obtained with WW, BY and BB perimetry were respectively: Reduction of sensitivity per year: 0.13, 0.27 and 0.13 dB; correlation coefficient (r) of threshold with age (and error of estimation of Y over X): -0.63 (2.24 dB), -0.70 (3.77 dB) and -0.80 (1.32 dB); threshold fluctuation: 2.21, 3.03 and 2.03 dB; percentage of points deviated more than 5dB from the expected value for the patient age: 8.1, 16.0 and 4.2%. CONCLUSIONS Perimetric results are more stable with BB strategy than with the other two types of perimetry. BY perimetry gives the worst results: threshold reduction with age is twice higher, individual fluctuation is 50% higher and points away from the mean value are much more frequent. Overlapping between blue and yellow filters is minimal in Octopus. Therefore, an absolute threshold is examined, which is much more unstable than WW or BB differential thresholds.