Sun sensitivity is a major risk factor for melanoma, basal cell carcinoma, and squamous cell carcinoma of the skin. Several variables have been used in epidemiologic studies to measure sun sensitivity. The present study assesses their validity and combines them to form a prediction rule for an objective measure of sun sensitivity, the minimal erythema dose of ultraviolet B radiation required to produce visibility reddened skin (MED). Participants were 116 patients with psoriasis presenting for phototherapy who completed a sun sensitivity questionnaire. Of the 14 questionnaire items evaluated, 10 were associated with the MED beyond expectation based on chance. The closest association was with the skin type (of Fitzpatrick), a 4-point scale based on historical ability to tan and susceptibility to sunburn. Color of untanned skin and hair were also independent predictors, and were included in the final prediction rule, which correlated 0.55 with MED. Combining items yields a more accurate predictor of sun sensitivity than any one or two individual response variables, and hence may be preferable for epidemiologic studies.