Image-guided systems are becoming more widely used in endoscopic sinus and skull base surgery. All systems require initial registration to correlate the CT scan images to the patient's anatomy. Multiple registration techniques can be used. The ideal technique is one that is easy, reproducible, and provides the most accurate registration in the least amount of time. This study used an optical-based image-guided system (LandmarX) to test a unique mask registration technique and (compared it to a previously used anatomic registration technique. Twenty-one patients were scanned with the mask and underwent surgery. Registration was performed using both the registration mask and the anatomic landmarks. Mean registration error and time were recorded. Results are reported for 20 patients. Mean registration error for the mask technique was 0.96 mm and took a mean of 41 seconds. Anatomic registration error using five or six points resulted in a mean initial error of 2.08 mm and took 31.2 seconds. Mean final anatomic registration error was 1.53 mm, requiring reregistration of a mean of 4.6 points, and took 106 seconds. Statistically significant differences were obtained between the two techniques with regard to registration error and time to final registration. We found that the registration mask technique is a more reliable technique in ease, accuracy, and time of registration. This technique should be especially beneficial to the less experienced image-guided surgeon.