Berlin is a melting pot of cultures, its residents coming from more than 160 different countries and bringing their heritage languages with them. In our X-Tutorial Metrolingualism in Berlin, we investigated whether speakers of heritage languages are likely to pass these on to the next generation.  As part of our study, we conducted 30 interviews with speakers of 35 mostly non-European heritage languages, asking about language use in their families, at work and in other contexts. We found that some communities were significantly less likely to maintain their heritage languages, resulting in lower language vitality.  Several models for assessing language vitality, among them EGIDS, AES, LEI (ELCat) and the UNESCO Atlas, have been proposed so far. All these models predict language vitality on a global scale. Our research shows, however, that their applicability to urban contexts is limited.  In this talk, we present our own approach to assessing urban language vitality based on the speakers’ responses in the interviews. We consider existing models and discuss how they can be adapted to fit urban contexts, taking into account factors such as the existence of ethnic neighborhoods, involvement of embassies and community size and cohesion. We conclude by presenting a preliminary proposal for a new model which accounts for these factors.