Various factors influencing the resolution limit (RL) in electron beam (EB) lithography have been analyzed quantitatively using the edge roughness evaluation (ERE) method. The ERE method is based on the experimental finding that line edge roughness of a resist pattern is inversely proportional to the slope of the Gaussian-distributed quasi-beam-profile (QBP). The analysis reveals that beam blur and development process are primary factors of RL, that electron forward scattering is negligible with the resist thickness under 200 nm, and that the effect of aperture degradation on RL is as large as that of resist performance. A necessary and sufficient condition for realizing 50 nm patterns is following. Use of EB lithography instruments of which beam blur is under 31 nm, resist thickness of 200 nm, and 20% improvement of the resist performance accompanying optimum development condition.