Human cancer risks are inversely correlated with (a) blood retinol and (b) dietary beta-carotene. Although retinol in the blood might well be truly protective, this would be of little immediate value without discovery of the important external determinants of blood retinol which (in developed countries) do not include dietary retinol or beta-carotene. If dietary beta-carotene is truly protective--which could be tested by controlled trials--there are a number of theoretical mechanisms whereby it might act, some of which do not directly involve its 'provitamin A' activity.