The subjects in the study were children who were X-rayed because of increased risk for resorption following ectopically erupting maxillary canines. One hundred and seven children 9 to 15 years of age with 156 maxillary canines that were erupting ectopically and 58 normally were investigated by computed tomography (CT) to describe the features of the dental follicles of the erupting maxillary canines. Contiguous, transverse CT scans were exposed through the maxilla in the canine region and the width and shape of the dental follicles were registered scan by scan throughout the extension of the follicle. The width and the shape of the dental follicle of the erupting maxillary canine varied greatly. The range of the maximum width, measured from the crown to the periphery of the follicle, was 0.5-7.0 mm, with a mean of 2.9 mm and a 95% confidence interval of 2.7-3.2 mm for the entire sample. No relationship was found between the width or shape of the follicles and sex, age, stage of eruption, inclination of the canine, or width of the dental arch. However, the location of the maxillary canine vis-à-vis the adjacent incisor was significantly associated with the width of the follicle, which indicated that local anatomic conditions might influence the width and shape of the follicle. The dental follicles of the ectopically erupting canines were, on average, wider than those of the normally erupting canines. The 95% confidence interval for the normally erupting canines was 2.3-2.7 mm; for the buccally erupting canines 2.4-4.1 mm; for the lingually erupting canines 2.6-3.0 mm; and for the apically erupting canines in relation to the lateral incisors 2.9-4.1 mm. Canine follicles that were wide but within normal limits did not cause deviations in adjacent teeth. Cystically degenerated dental follicles were found but were indistinguishable on the CT scans from those that had been widened physiologically. The contributions of the studied variables to the variation in the width of the dental follicle of the maxillary canine were analyzed with regression models. [ABSTRACT FROM AUTHOR]