It has been consistently reported that young males commit crimes with an average frequency five times greater than their female peers. Most data supporting this view are derived from juvenile court and police statistics. Studies using data derived from self-reported behaviour suggest that the true relative frequency may be closer to 2:1. Police and juvenile justice data for the year 1994-5 in New South Wales, Australia, were analysed in an attempt to determine whether court and police statistics might reflect a form of selection bias, where the likelihood of arrest, trial and/or sentence is a function of gender, rather than frequency and nature of offence. The results suggest that the 5:1 gender ratio reflects a strong component of gender bias in the workings of the juvenile justice system in New South Wales. If suspected of a given crime, young males are more likely to be denied bail and (if found guilty) to be given a harsher sentence than young females suspected (or found guilty) of the same crime. Overall, if found guilty of an offence, boys were four times more likely than girls to receive a custodial sentence. Therefore, boys are selectively denied access to alternate rehabilitation resources which are made available to girls who are in trouble with the law. The juvenile justice system in New South Wales requires careful examination and reform if such apparently deeply entrenched biases are to be eliminated.