We compared the true osteoid seam width (TOSW) as measured by a modification of the orthogonal intercept lengths with various methods of estimating seam widths, including (1) the commonly used length measurements at four equidistant points (O.Wi/4PT), (2) osteoid area divided by the osteoid perimeter (O.Ar/O/Pm) or the bone/osteoid interface (O.Ar/B.Bd), and (3) a novel method for estimating seam width defined as the osteoid area divided by the major axis of the seam (O.Ar/Axis). All methods for approximating osteoid seam width significantly exaggerated the true osteoid seam dimension by an amount that ranged from 16 to 23%. However, the relative accuracy of all methods of estimating osteoid seam width are equivalent as evidenced by the similar mean difference from the TOSW (3.4, 4.1, 5.1, and 3.8) demonstrated by O.Wi/5PT, O.Ar/Axis, O.Ar/O.Pm, O.Ar/B.Bd, respectively. Regression analysis of the various estimates of seam width on TOSW also demonstrated the equivalency of these methods. Moreover, all estimates could be employed to discriminate seams of normal dimensions from abnormally wide seams in bone specimens derived from patients with osteomalacia. Differences between the methods, however, were observed that may have practical importance. In this regard, the direct procedure of determining distance demonstrated less variance than the indirect estimate of width. As a result, the direct measurement required fewer samples (n = 13) to detect a significant difference to normal and could discriminate smaller deviations in seam width (1.7 microns) at a given sample size compared with O.Ar/Axis (n = 28; 2.9 microns), O.Ar/O.Pm (n = 42; 3.4 microns), and O.Ar/B.Bd (n = 42; n = 3.2).(ABSTRACT TRUNCATED AT 250 WORDS)