This paper outlines the elements of a combined experimental/analytical methodology for accurate shear characterization of unidirectional composites in the linear and nonlinear range, with particular attention devoted to metal matrix composites. It is illustrated that consistent results can be obtained for a large class of composites from two commonly employed shear test methods currently in use by composites researchers when the influence of various factors that affect the determination of the actual shear response is properly analyzed. These factors include the effects of material anisotropy, specimen geometry, manner of load introduction, and test fixture design on the stress and deformation fields in the test section of off-axis and losipescu specimens. Common errors associated with the measurement of deformation fields and calculation of stress fields are discussed and quantified. Particular problems in the determination of the shear response of unidirectional boron/aluminum using the losipescu test are illustrated and discussed.