Many prediction guidelines exist in facial approximation for determining the soft-tissue features of the face, and the reliability of each is generally unknown. This study examines four published and commonly used soft-tissue prediction guidelines for estimating nose projection, two of which also estimate the position of the pronasale. The methods tested are those described by: 1) Gerasimov ( The Face Finder; London: Hutchinson & Co.), using the distal third of the nasal bones and the nasal spine; 2) Krogman ( The Human Skeleton in Forensic Medicine; Springfield: Charles C. Thomas), using the average soft-tissue depth at midphiltrum, plus three times the length of the nasal spine (and a variation of this technique: plus three times the distance of the tip of the nasal spine from the nasal aperture); 3) Prokopec and Ubelaker ( Forensic Sci Commun 4:1–4), using the reflected profile line of the nasal aperture; and 4) George ( J Forensic Sci 32:1305–1330), using a variation of the Goode method. Four identical hard-tissue tracings were made of 59 adult lateral head cephlograms (29 males, mean age 24, SD 10 years; 30 females, mean age 23, SD 5 years) on separate sheets of tracing paper. One soft-tissue tracing was also made for each radiograph. All tracings were marked with three identical reference points. Soft-tissue tracings were isolated from one of us (C.N.S.), who attempted under blind conditions to predict pronasale position and nose projection on the hard-tissue tracings, using the soft-tissue prediction guides above. Actual soft-tissue tracings were then compared to each of the predicted tracings, and differences in projection/pronasale position were measured. Results indicate that for nose projection, methods 3 and 4 performed well, while methods 1 and 2 performed poorly. Features which are most related to nose projection/pronasale are described in this paper, as are regression equations generated from these variables that predict pronasale/nose projection better than the traditional methods mentioned above. The results of this study are significant because they: 1) indicate that the popular facial approximation methods used to build the nose are inaccurate and produce incorrect nose anatomy; and 2) indicate that the new pronasale prediction methods developed here appear to have less error than traditional methods. Am J Phys Anthropol, 2003. © 2003 Wiley-Liss, Inc.