Metal building systems are widely used in low-rise (1- or 2-story) building construction for economic reasons. Maximum cost efficiency is usually achieved through optimization of steel weight and the fabrication process by adopting web-tapered members and bolted end-plate connections. However, the cyclic behavior of this kind of system has not been investigated, and no specific seismic design guidelines are available in the United States. Based on both experimental and analytical studies, this dissertation introduces a new design concept utilizing drift evaluation, and proposes a seismic design procedure for metal building systems. Full-scale cyclic testing on a metal building with web-tapered members demonstrated that the system has high deformability, but little ductility. Proper flange bracing was essential to prevent premature lateral-torsional buckling. Test results also showed that the overstrength of this system was very high since the non-seismic load combination governed the design. A correlation study indicated that the failure modes corresponded well with the strength evaluation contained in the AISC LRFD Specification. Numerical simulation using the finite element analysis program ABAQUS demonstrated that good correlations in both the failure mode and the system strength characteristics could be achieved when a proper assumption on the initial geometric imperfections was made. A parametric study showed that the best correlation was found in the models with the first eigen buckling mode shape and an amplitude of Lb/1000 as an initial imperfection (Lb = unbraced length). A drift-based seismic design procedure was then developed. The design goal is to ensure that the elastic drift capacity of the system is larger than the drift demand with a sufficient margin. The drift capacity is calculated using the system overstrength factor, and both drift capacity and demand are estimated utilizing the actual fundamental period. The proposed factor of safety (= 1.4) partially reflects the influence of low damping nature of metal buildings. Case studies using the proposed design procedure indicated that metal frames with heavy walls (i.e., masonry or concrete) based on the current design procedure are very vulnerable to collapse under major earthquake events, but the impact to the design of typical metal buildings without heavy wall attachments is insignificant