Mechanisms proposed to explain efficient melt transport away from hot, ductile source regions are problematic. Brittle-elastic fracturing is a well-known mechanism that allows fast magma migration as dikes through cold crust. Ductile fractures have been propo sed as an alternative for ductile environments, where brittle-elastic diking is inhibited. Ductile fracturing results from rock creep and growth of microscale voids that become interconnected, leading to rock failure. In this paper, we present observations and numerical models supporting the hypothesis that ductile fracture controls early steps in magma migration. We postulate that once developed, ductile fracture dikes may reach a critical length where magma stresses at dike tips overcome fracture toughness and lead to brittle-elastic diking, which subsequently controls magma migration.