Previous studies have shown that behavior of parents is associated with the criminal behavior of their offspring. Those with criminal or imprisoned parents have been shown to be more likely to become offenders themselves compared to their peers with non-criminal parents, while also parental divorce and early motherhood are risks factor for the development of antisocial and delinquent behavior in offspring. These studies, however, usually do not control for unmeasured confounders and these associations might therefore be spurious. In this study, a children of siblings design was used in which the criminal behavior of children of discordant siblings (i.e., one criminal father and his non-criminal brother) was compared. This model controls (partly) for confounding by genetic and shared environmental factors and therefore gives a better estimate of the causal effects. Using register data of Statistic Netherlands all children born in the Netherlands between 1991 and 2001 were selected for analyses on the population level (N&2,100,000). Next, the offspring of siblings who were discordant on parental divorce, paternal offending, and maternal age at birth were compared in three separate conditional logistic regression analyses. The results show that the intergenerational associations become considerably smaller or even disappear when a children of sibling design is used compared to when unrelated individuals are compared. This indicates that these associations are at least partly confounded by genetic or shared environmental factors, and that intergenerational associations are overestimated when these confounding factors are not controlled for.