The aim of this study is to advance scholarship on the IQ–offending relationship by examining the functional form of this relationship and whether confounding introduced by socioeconomic status (SES) and other factors can be adequately addressed. Data from the National Longitudinal Survey of Youth are analyzed using generalized propensity score and propensity score matching analyses. The results suggest that the relationship is curvilinear, such that lower and higher levels of IQ are associated with lower levels of offending. They also indicate that the distribution of confounders, especially SES, may limit the ability of statistical approaches to arrive at unbiased estimates of IQ effects.