The humerus and femur of the fossil hominid OH 62 are badly damaged and their lengths are not directly measurable (Johanson et al., 1987). Nevertheless, using relatively intact reference materials from another early hominid, AL 288-1, Johanson et al. (1987) reconstructed the bones to estimate the humerofemoral index, which falls well above the range for modern Homo, above the estimate for AL 288-1, and within the range for Pan paniscus. The reconstruction of missing bone by the method originally employed for OH 62 is broadly reproducible in a representative modern sample of Homo, making possible the estimation of an associated error term intrinsic to this method. Using the approximate variance of the ratio mean (Kish, 1965), shown here to be a good estimator of the sample variance of the humerofemoral index, the analysis of this modern sample extrapolated to other living hominoids gives quite acceptable results. Applied to OH 62, it suggests an error term associated with the estimated humerofemoral index so substantial that it is only possible to situate the index somewhere between the distributions for Homo and Gorilla, and quite possibly not above the index for AL 288-1. On the other hand, the predicted distribution for the humerofemoral index of AL 288-1 is more securely placed between the distributions for Homo and Pan paniscus.