The procedural justice model of legal socialization holds that personal fairness judgments influence criminal offending directly and indirectly, via legal orientations (e.g. legitimacy and legal cynicism). This study used longitudinal data to empirically scrutinize these arguments. Using 11 waves of data from the Pathways to Desistance study (i.e. baseline and 10 follow-up interviews), a series of time-lagged, multi-level longitudinal regression models with time-varying and time-stable measures decomposed into between- and within-individual components were estimated. The estimates from the linear mixed-effects models showed that procedural justice judgments directly influence legitimacy and, though relatively more limited, legal cynicism over time both between- and within-individuals. Test statistics indicated too that positive procedural justice judgments reduce involvement in criminal offending between individuals. However, legitimacy is found to significantly mediate the effect of personal procedural justice judgments. The effect of vicarious procedural justice judgments remained statistically significant in explaining differences in criminal offending. As for the within-individual model, neither procedural justice scale predicted offending. However, legal cynicism did have a direct significant effect in the within-individual offending model. The findings support key elements legal socialization theory, especially the premise that the way in which criminal justice authorities treat the individuals they come into contact with influences (albeit indirectly) subsequent compliance with the law.