Controlling the microstructure of materials by means of phase separation is a versatile tool for optimizing material properties. In this study, we show that ink jet 3D printing of polymer blends gives rise to controllable phase separation that can be used to tailor the release of drugs. We predicted phase separation using high throughput screening combined with a model based on the Flory-Huggins interaction parameter, and were able to show that drug release from 3D printed structures can be predicted from observations based on single drops of mixtures. This new understanding gives us hierarchical compositional control, from droplet to device, allowing release to be ‘dialed up’ without any manipulation of geometry. This is an important advance for implants that need to be delivered by cannula, where the shape is highly constrained and thus the usual geometrical freedoms associated with 3D printing cannot be exploited, bringing a hitherto unseen level of understanding to emergent material properties of 3D printing.