This article presents an approach for measuring the educationally relevant writing ability of secondary school students, as applied in the project Multilingual Development: A Longitudinal Perspective (MEZ). Due to the longitudinal design of the study, the aspect of competence measurement across time is essential. The participants were adolescent students with German-Russian, German-Turkish, and monolingual German language backgrounds. In four waves, they were tested in German (majority language), Russian or Turkish (languages of origin) as well as English, French and/or Russian (foreign languages at school). We present a cross-lingual, comprehensive competence model of writing skills and its operationalization through empirical indicators. Subsequent statistical modelling using a longitudinal latent measurement model allows the explicit testing of measurement characteristics such as measurement invariance. The results reveal that the postulated theoretical construct has been measured in all languages in a structurally comparable way, and the development of inter-individual differences over time can be reliably interpreted. However, the interpretation of intra-individual developmental patterns over time is not yet possible.