A population-based prevalence survey of skin cancer was conducted in Nambour, Queensland, in 1986. The skin colour of 807 participants was assessed in three ways: quantitatively, graded by a dermatologist, and self-reported. Quantitation of skin pigmentation was obtained by measuring the reflectance of light of wavelength 650 nm, at six sites. Females showed higher mean reflectance (paler skin) than males at all sites with the greatest difference on the lateral forearms. Prevalent skin cancer in males, and solar keratoses in both sexes were correlated with inherently pale skin colour on an unexposed site, and the presence of keratoses was correlated with darkly-pigmented backs of the hands (P less than 0.001). Both dermatologists' and participants' grading of skin colour were moderately correlated with measured skin colour. For dermatologists, correlation was highest with reflectance from the medial upper arms (r = 0.35, right arm; 0.30, left) in males, and the lateral forearms (r = 0.34, right; 0.38, left) in females. Correlations between reflectance values and self-reported innate skin colour were highest for the upper arms (r = 0.26, right; 0.24, left) in males, and for forearms (r = 0.42, right and left) in females. Prevalence of actinic lesions was more highly correlated with subjectively assessed skin colour than with quantitative skin pigmentation.