Language endangerment is by now a well-known topic, and language documentation one well-established academic discipline that aims at addressing this urgent issue. Language documentation in the modern sense is concerned with creating lasting records of language in the natural environment by building annotated multi-media corpora, among other resources. A crucial point here is to ensure that the data are archived in a sustainable way – they ought to be available and usable for years and decades to come, as the basis for further research, educational projects or language revitalisation activities. Generally there is yet little awareness of the fact that the data about endangered languages are endangered themselves. This talk presents the activities and solutions being developed at The Language Archive at the Max-Planck-Institute for Psycholinguistics in Nijmegen that aim at providing tools and an infrastructure that supports the creation and long-term archiving of precious language data.