Implant placement in the posterior maxilla often requires elevation of the sinus floor, which can be achieved through either the modified Caldwell-Luc or the crestal osteotome technique. The objectives of this study were to evaluate (a) the resistance to perforation of maxillary sinus membranes obtained from formaldehyde-fixed cadavers in vitro, (b) the frequency and extent of membrane perforations occurring after sinus floor elevation in cadavers using the crestal approach, and (c) the amount of membrane elevation (doming) that can be achieved using the crestal approach. Pretreatment of maxillary sinus membrane tissues with commonly used tissue softeners did not have a statistically significant effect on resistance to perforation. Maxillary sinus membranes were elevated 4 to 8 mm in formaldehyde-fixed cadavers using the osteotome technique; implants were placed. Of the 25 sites that received implants, only 6 showed perforations, as assessed by double-blind investigation after dissection of the lateral wall of the nose, allowing direct examination of the sinus cavity. Perforations were categorized as Class I (< or = 2 mm with exposure of the implant into the sinus cavity and loss of doming); Class II perforations (> or = 2 mm) were associated with proximity of the osteotomy site to the medial wall of the sinus or the presence of septae. These results indicated that the crestal osteotome approach compared favorably to the modified Caldwell-Luc technique as it relates to the frequency of maxillary sinus membrane perforations and the degree of achievable membrane elevation.