Opioid mortality increases have been linked to both lax and restrictive opioid prescription regulations. Modeling choice between prescription and illicitly manufactured opioid sources helps reconcile the apparently contradictory empirical findings. It also identifies groups responding opposite of the average and applies previous studies to new supply conditions. Organized around the two supply channels, a policy database is assembled that reveals distinct pricing phases during 1999-2021. Consistent with the model, during the later phases the relationship between the opioid fatality rate (measured from death certificates) and its composition changes sign, minors' fatality rates trend opposite of adults', and the black-white gap changes sign.