Orthogonal intercepts from random test lines (OIr), uniformly distributed intercepts (OIu), and areas and lengths (Ar/Le) were measured manually and by computer to determine individual profile and sample widths of artificial profiles and human osteoid seams. Individual widths were equal by Ar/Le and OIu methods. The means of individual profile widths (all methods), of all orthogonal intercepts (OIu and OIr), or of profile widths weighted in proportion to their lengths (Ar/Le) were also equal. Ar/Le and OIu had smaller variance than OIr. Discrepant individual OIu and Ar/Le widths in digitized images were corrected by thresholding and did not significantly affect sample means. Unweighted Ar/Le sample means were 15-44% lower than weighted means. Distributions of osteoid seam widths were not normal, and all but one had more than one mode. We conclude that the Ar/Le method is comparable to direct orthogonal intercept methods and suitable for automated histomorphometry.