In recent years, the spreading of visual augmented reality as an effective tool in image-guided surgery, has stimulated the research community to investigate the use of commercial augmented reality headsets a broad range of potential applications. This aroused enthusiasm among clinicians for the potential of augmented reality, but also revealed some technological and human-factor limitations that still hinder its routine adoption in the operating room. In this work, we propose an alternative to head-mounted displays, based on projected augmented reality. Projected augmented reality completely preserves the surgeon's natural view of the operating field, because it requires no perspective conversion and/or optical mediation. We selected a cranio-maxillofacial surgery application as a benchmark to test the proposed system and compare its accuracy with the one obtained with a video see-through system. The augmented reality overlay accuracy was evaluated by measuring the distance between a virtual osteotomy line and its real counterpart. The experimental tests showed that the accuracy of the two augmented reality modes is similar, with a median error discrepancy of about 0.3 mm for the projected augmented reality mode. Results suggest that projected augmented reality can be a valuable alternative to standard see-through head-mounted displays to support in-situ visualization of medical imaging data as surgical guidance. INDEX TERMS augmented reality, projected augmented reality, video see-through, head-mounted display, computer-assisted surgery, cranio-maxillofacial surgery, augmented reality accuracy