The floral ontogeny of Neurada procumbens L. has been investigated to clarify the position of the family Neuradaceae and to elucidate a number of morphological problems, such as the nature of the epicalyx and the 10-carpellate gynoecium. Morphological and ontogenetic evidence suggests a close affinity with the Rosaceae, contrary to reports of rbcL sequence data. There is a strong gynoecial resemblance with Maloideae, but other characteristics suggest that the affinities of Neuradaceae lie at the base of the Rosaceae.