Experimental study of the in vivo kinetics of tanning in human skin has been limited by the difficulties in measuring changes in melanin pigmentation independent of the ultravioletinduced changes in erythema. The present study attempted to experimentally circumvent this issue. We have studied erythemal and tanning responses following a single exposure to a range of doses of ultraviolet B irradiation on the buttock and the lower back in 98 subjects. Erythema was assessed using reflectance techniques at 24 h and tanning measured as the L* spectrophotometric score at 7 days following noradrenaline iontophoresis. We show that dose (P < 0.0001), body site (P < 0.0001), skin colour (P < 0.0001), ancestry (P = 0.0074), phototype (P = 0.0019) and sex (P = 0.04) are all independent predictors of erythema. Quantitative estimates of the effects of these variables are reported, but the effects of ancestry and phototype do not appear solely explainable in terms of L* score. Dose (P < 0.0001), body site (P < 0.0001) and skin colour (P = 0.0365) or, as an alternative to skin colour, skin type (P = 0.0193) predict tanning, with those with lighter skin tanning slightly more to a defined UVB dose. If erythema is factored into the regression, then only dose and body site remain significant predictors of tanning: therefore neither phototype nor pigmentary factors, such as baseline skin colour, or eye or hair colour, predict change in skin colour to a unit erythemal response.