We measured non-haem Fe absorption with and without added Ca in a short-term feeding study, in thirteen women with marginal Fe status, by the use of a double stable isotope technique. Supplementing 500 mg Ca as calcium carbonate significantly (P = 0·0009) reduced Fe absorption from a single meal from 10·2 % (range 2·2–40·6) to 4·8 % (range 0·7–18·9). A significant inverse correlation in the absence ( − 0·67, P = 0·010) and presence ( − 0·58, P = 0·037) of Ca, respectively, was found between Fe absorption and Fe stores measured by serum ferritin (SF). Wide variation in Fe absorption was observed between individuals in the absence and in the presence of Ca, despite pre-selection of participants within a relatively narrow range of iron stores (SF concentrations). Correction of Fe absorption data based on group mean SF was not found to be useful in reducing the inter-individual variability in iron absorption. It appears that selecting a study group with a narrow initial range of Fe stores does not necessarily reduce the inter-individual variability in Fe bioavailability measurements. These results support the hypothesis that body Fe stores, although an important determinant of dietary Fe absorption, are not the main factor that determines Fe absorption under conditions of identical dietary intake in subjects with low Fe stores.