Voice assimilation in Hungarian: the hitches Voice assimilation is a much discussed issue recently. The topics involved are whether the feature voiced is privative or equipollent and whether the process can be described by making reference to prosodic structure or by specifying the environment linearly. By analysing voice assimilation in Hungarian, I attempt to show that a privative feature is capable of doing the job -- though not unproblematically -- and that the environment is much more easy to describe simply by making reference to the next segment(s) than by identifying the position of the target and the trigger by traditional syllabic constituents, like onset and coda. The last section of the paper treats three segments, [v h j], the ambivalent behaviour of which raises a number of problems to be solved. The aim is not so much to provide solutions to each problematic detail, but to gather the points that have to be sorted out.