Abstract
Recent work by Sandra L. Huffman and her colleagues on the determinants of the length of amenorrhoea in Bangladesh is examined. It is shown that the data on which their work is based are contaminated by an important period effect: the 1974–75 famine in Bangladesh. The famine is described and its possible effects on Huffman's results examined. Huffman argues that breastfeeding behaviour is a more important factor in prolonging amenorrhoea among lactating women than is the mother's nutritional status. It is shown, however, that measures for the key variables used by Huffman - nutritional status, breastfeeding behaviour, and length of amenorrhoea - were all affected by the famine, and that the famine effects may have biased her findings. This does not necessarily mean that her hypothesis is wrong, but does suggest that it must be tested in unbiased data.