Previous epidemiological studies conducted in retinol-supplemented subjects showed an association between high serum levels or dietary intake of retinol and risk of hip fracture. On the other side, observational studies revealed that non-supplemented subjects with higher dietary intake of retinol lose less bone with age than subjects with lower intake. This discrepancy, currently unexplained, suggests that nutrition plays a major role in conditioning the effects of retinol on bone. Since retinol is derived from both retinoids--contained in animal food--and carotenoids--contained in vegetables and fruits--we evaluated a possible role of carotenoids in involutional osteoporosis. Therefore, plasma levels of beta-carotene and other carotenoids, in addition to those of retinol, were measured in free-living, non-supplemented, elderly women with or without severe osteoporosis. Plasma levels of retinol and of all carotenoids tested, with the exception of lutein, were consistently lower in osteoporotic than in control women. A weak association was found only between retinol and femoral neck bone mineral density in osteoporotic women. Our study suggests a bone sparing effect of retinol, to which the provitamin A activity of some carotenoids might have contributed.