The glandular odontogenic cyst (GOC) is now a relatively well-known entity with recent reviews indicating over 100 cases reported in the English literature. The GOC's importance relates to the fact that it exhibits a propensity for recurrence similar to the odontogenic keratocyst, and that it may be confused microscopically with central mucoepidermoid carcinoma (CMEC). Numerous histopathologic features for the GOC have been described, but the exact microscopic criteria necessary for diagnosis have not been universally accepted. Furthermore, some of the microscopic features of GOC may also be found in dentigerous, botryoid, radicular, and surgical ciliated cysts. The purpose of this multicenter retrospective study is to further define the clinical, radiographic, and microscopic features of GOC, to determine which microscopic features may be helpful for diagnosis in problematic cases, to determine the most appropriate treatment, and to determine if GOC and CMEC share a histopathologic spectrum. In our series of 46 cases, the mean age at diagnosis was 51years with 71% of cases in the 5th-7th decades. No gender predilection was noted. 80% of cases occurred in the mandible, and 60% of the lesions involved the anterior regions of the jaws. Swelling/expansion was the most common presenting complaint, although some cases were asymptomatic. Radiographically, most cases presented as a well-defined unilocular or multilocular radiolucency involving the periapical area of multiple teeth. Some lesions displayed a scalloped border. Cases also presented in dentigerous, lateral periodontal, and "globulomaxillary" relationships. The canine area was a common location for maxillary cases. All cases were treated conservatively (enucleation, curettage, cystectomy, excision). Follow-up on 18 cases revealed a recurrence rate of 50% (9/18), with 6 cases recurring more than once (range of follow-up: 2months to 20years; average length of follow-up: 8.75years). The mean interval from initial treatment to first recurrence was 8years, and from first recurrence to second recurrence was 5.8years. Two cases recurred three times and the interval from second to third recurrence was 7years (exact interval only documented in one case). All cases exhibited eosinophilic cuboidal (hobnail) cells, a feature not specific for GOC, but necessary for diagnosis, in our opinion. Univariate analysis indicated several features that are most helpful in distinguishing GOC from GOC mimickers in problematic cases, including: (1) the presence of microcysts (P<0.0001); (2) epithelial spheres (P<0.0001); (3) clear cells (P=0.0002); (4) variable thickness of the epithelial cyst lining (P=0.0002); and (5) multiple compartments (P=0.006). Stratified analysis indicated that when microcysts are present, epithelial spheres and multiple compartments are still significant, and clear cells are marginally significant in distinguishing GOCs from GOC mimickers. The presence of microcysts (P=0.001), clear cells (P=0.032), and epithelial spheres (P=0.042) appeared to be most helpful in distinguishing GOC associated with an unerupted tooth from dentigerous cyst with metaplastic changes. There were no statistically significant differences microscopically between GOCs that recurred and those that did not. The presence of 7 or more microscopic parameters was highly predictive of a diagnosis of GOC in our series (P<0.0001), while the presence of 5 or less microscopic parameters was highly predictive of a non-GOC diagnosis (P<0.0001). Islands resembling mucoepidermoid carcinoma (MEC-like islands) were identified in the cyst wall of three cases, only one of which had follow-up (no evidence of disease at 74 mo.); therefore, at this time insufficient information is available to determine whether GOC and CMEC share a histopathologic spectrum or whether MEC-like islands in GOCs are associated with more aggressive or malignant behavior.