Recently, Michael Gottfredson and Travis Hirschi have dismissed longitudinal studies in criminology as pointless and unduly expensive. They state their case in quite general terms but have a narrow and unusual definition of longitudinal studies. Their arguments refer to one type of such studies, i.e., the protracted prospective study with data collected in waves, whereas their preferred alternative, the “cross-sectional survey” can include another type, the retrospective study with retrospective data. After a discussion of the extent to which individual-level change can be empirically analyzed with repeated cross-sectional data in the strict sense, it is concluded that such data permit only inefficient or highly special analysis. Gottfredson’s and Hirschi’s own general theory of crime claims differential constancies for basic variables. It is found that an empirical test of this central part of their theory cannot be made on the basis of a cross-sectional survey but requires a protracted prospective longitudinal study.