The lack of adequate documentation for African languages is a major challenge facing linguists. Understandably, priority has often been given to the most endangered languages, but the level of language endangerment in Africa has been grossly underestimated. Language shift can occur in a single generation, so population surveys that are 20 or more years old cannot be used to label a language “safe,” particularly when it is used by only a few thousand people. High rates of language shift are being reported in different parts of the continent, and even larger languages (with 100,000 or more speakers) might best be considered threatened. Documenting an obsolescent language is a difficult task, made even more difficult if the language is essentially undescribed. Since basic phonological and grammatical sketches are lacking for so many African languages, we should try to address the challenge of documenting these poorly known languages while they are still used by all generations.