The age at which girls tend to marry is one of the most important factors in determining the overall rates of fertility in a given population, and hence its general demographic profile. It also affects a whole range of social institutions of reproduction, above all the ‘shape’ of the family, the relationships between the mother and her children, between husband and wife, and the ways in which property can be redistributed through inheritance. It is the simple and restricted purpose of this paper to re-examine the data that have hitherto been used to determine the age at marriage of girls in Roman society. For the purposes of this study, ‘Roman society’ is defined as the conglomerate of urban-centred communities that developed in Europe west of the Adriatic, as well as in the lands of the Danubian Basin. It is conceded that family types and modes of family formation in the eastern parts of the empire were different from those in the west, and therefore require separate analysis. In performing this task, the analysis presented here also attempts to demonstrate the highly specific nature of the set of data employed in the ‘age-at-marriage’ debate, and to question its relevance to the age at first marriage of most girls in the western Roman empire. Having demonstrated the limited validity of these data, I shall then suggest another method that might usefully be employed to approach the problem. Finally, to complete the argument, a series of hypotheses will be advanced that seek to link the range and modes of age at first marriage of girls of various status groups and classes to other social and economic factors in the Roman world.