Following critiques that the DSM multiaxial system lacks psychodynamic information useful for treatment, an axis for defense mechanisms was developed for DSM-IV, including up to 7 individual defenses from a glossary of 27, and 3 predominant defense levels from a list of 7. We tested the feasibility, reliability, and discriminability of the proposed axis. Clinician and psychiatric resident volunteers were trained at two U.S. and one Norwegian sites. After conducting initial interviews on 107 patients, they rated the DSM-III-R and defense axes, as did a second blind rater. Median kappa reliabilities were .42 (individual defenses), and .47 (defense levels). A summary measure, Overall Defensive Functioning (ODF), had similar reliability to current GAF (IR .68 vs. .62), similar 1-month stability (.75 vs. .78), but greater 6-month stability (.51 vs. .17). Independent of Axis III, ODF had small to moderate associations with other Axes and symptoms. Our findings indicate that the defense axis is a feasible, acceptably reliable, and nonredundant addition to DSM-IV, which may prove useful for planning and conducting treatment.