It is unclear how well surrogate markers for vitamin D exposure (eg, oral intake of vitamin D and estimates of sunlight exposure), with and without consideration of other potential predictors of 25-hydroxyvitamin D [25(OH)D] concentrations, similarly rank individuals with respect to 25(OH)D blood concentrations.The objective was to determine how much variation in serum 25(OH)D concentrations (nmol/L) could be explained by a predictive model with the use of different vitamin D surrogate markers (latitude of residence, mean annual regional solar irradiance estimates, and oral sources) and other individual characteristics that might influence vitamin D status.A random sample of 3055 postmenopausal women (aged 50-70 y) participating in 3 nested case-control studies of the Women's Health Initiative Calcium plus Vitamin D Clinical Trial was used. Serum 25(OH)D values, assessed at year 1 (1995-2000), and potential predictors of 25(OH)D concentrations, assessed at year 1 or Women's Health Initiative baseline (1993-1998), were used.More than half of the women (57.1%) had deficient (<50 nmol/L) concentrations of 25(OH)D. Distributions of 25(OH)D concentrations by level of latitude of residence, mean annual regional solar irradiance, and intake of vitamin D varied considerably. The predictive model for 25(OH)D explained 21% of the variation in 25(OH)D concentrations. After adjustment for month of blood draw, breast cancer status, colorectal cancer status, fracture status, participation in the hormone therapy trial, and randomization to the dietary modification trial, the predictive model included total vitamin D intake from foods and supplements, waist circumference, recreational physical activity, race-ethnicity, regional solar irradiance, and age.Surrogate markers for 25(OH)D concentrations, although somewhat correlated, do not adequately reflect serum vitamin D measures. These markers and predictive models of blood 25(OH)D concentrations should not be given as much weight in epidemiologic studies of cancer risk.