Abstract: Together with experiments, the main method in the study of child language development is the analysis of behavioral changes of individual children over time. For this purpose recordings of infants’ and children’s naturalistic interactions in a variety of languages spoken in different cultural contexts are key. Language development corpora either are cross-sectional or longitudinal collections of conversations recorded at predetermined intervals and annotated on various linguistic and multi-modal levels. Here, we discuss the advantages and disadvantages of cross-sectional and longitudinal studies but the focus of this chapter will be on longitudinal studies which are the main tool in corpus analyses of child language. We focus on challenges of corpus design including sampling issues such as the number and choice of participants, amount of recordings, recording situations, transcription, linguistic and multi-modal annotations and ethical considerations.