A latent variable modeling approach to reliability and measurement invariance evaluation for multiple-component measuring instruments is outlined. An initial discussion deals with the limitations of coefficient alpha, a frequently used index of composite reliability. A widely and readily applicable structural modeling framework is next described that allows point and interval estimation of scale reliability. An easily employed procedure for testing measurement and reliability invariance in multiple population and longitudinal studies is then discussed. Unlike coefficient alpha that is in general a misestimator of scale reliability already at the population level, the article is based on the formal definition of the reliability coefficient, and its underlying modeling approach does not share alpha's limitations. The described reliability and measurement invariance evaluation procedures are illustrated on a set of data resulting from a Social Interaction Anxiety Scale, and source codes for one of the most popular structural modeling programs, LISREL, are provided, which can be used to apply the outlined methods.