The threshold of binocular depth perception was measured in 11 healthy volunteers. A three-rod arrangement was employed in which both the luminance of the rods and that of the adapting field could be adjusted independently. This allowed fixing the contrast when the effect of luminance was studied or fixing the luminance when the effect of contrast was investigated. The observation distance was 400 mm. Thresholds were expressed as angular disparities and were based on 75% correct responses. Points of subjective equality were also determined. Lowest thresholds (2.85 +/- 0.67 s of arc) were found for a moderate contrast of 0.5 whereas low (0.05) and high (0.95) contrast both produced significantly higher thresholds (luminance 250 cd/m2). Altering the field luminance (50, 250, 1600 cd/m2) under constant contrast conditions (0.95) did not measurably influence stereoscopic acuity.