Though less productive than rival word-formation processes like compounding and affixation, blending is still a rich source of neologisms in French. Despite this productivity, however, blends are often seen by scholars as unpredictable, uninteresting, or both. This analysis picks up where recent studies of blending have left off, using Correspondence Theory and a bundle of segmental constraints to deal with this phenomenon as it pertains to French. More specifically, it shows that blending is the result of a single output standing in correspondence with two or more other outputs, and that we do not need to refer to prosodic information, which is crucial in accounts of blending in languages with lexical stress like English, to account for the process in French. The analysis also differs from previous studies in that it locates blending exclusively within the phonology, leaving its morphological and semantic characteristics to be handled by other processes in the grammar.