Low self-control is at the heart of Gottfredson and Hirschi's (1990) general theory of crime and had been extensively investigated with the Grasmick, Tittle, Bursik, and Arneklev (1993) self-control scale. An advanced analytical tool, confirmatory techniques, had been recently applied to elucidate the underlying theoretical structure of the self-control construct. Most of these studies were limited,as they did not compare competing factorial measurement models. Furthermore, measures such as the Grasmick et al. scale had been used widely with diverse population groups and it was essential that the invariance of the measure be examined. The few studies that had performed this had addressed the invariance of the scale's factorial structure, but neglected to examine the invariance of the latent means structures. Testing the invariance of the latent means eliminates spurious results that may be artifacts of biased responding. The purpose of the current study was to investigate the structure of self-control by comparing three measurement models using confirmatory factor analysis and to explore the invariance of self-control across multiple groups from New Zealand. Two male groups were used: prison inmates and university students. The overall findings supported the Grasmick et al. scale as usefully represented by both a hierarchical order and a multidimensional structure. The measure was found to be generalizable across disparate groups,although differences in latent means were found between the two groups on the self-control dimensions.