The success of measurement — the quality of the inferences provided by a measurement model — is usually assessed with reference to two key concepts: validity and reliability. The distinct uses of measures of latent variables are reported. The article then deals with the costs of ignoring measurement error. Additionally, a quick introduction to factor analysis, item-response models, and a very general class of latent variable models are briefly given. Moreover, it describes the inference for discrete latent variables and the measurement in a dynamic setting.