One of the critical steps in high throughput crystallography that so far has evaded automation is the actual harvesting of the delicate crystals from the mother liquor in which they are growing. The late-stage operation of harvesting is presently a most risky and loss-intensive procedure, compounded by its tight integration with the critical steps of cryo-protection and cryo-quenching. Recent advances in micromanipulation robotics and micro-fabrication have made it possible to seriously consider automation of protein crystal harvesting. Based on the experience gained during the development of an operator-assisted (and now operator-assisting) universal micromanipulation robot (UMR) prototype, we discuss the challenges ahead for the design of a fully autonomous, integrated system capable of the reliable harvesting of protein microcrystals. Experience from participation in NIH structural genomics projects and feedback from bottleneck workshops indicates that genuine demand exists in the high throughput community as well as in pharmaceutical production pipelines, justifying the effort and resources to develop autonomous harvesting robotics.