Purpose: One reason why the diagnosis of apraxia of speech (AOS) and its underlying impairment are often debated may lie in the fact that most patients do not display pure patterns of AOS. Mixed patterns are clearly acknowledged at other levels of impairment (e.g., lexical-semantic and lexical-phonological), and they have contributed to debate about the degree of interaction between encoding levels; by contrast, mixed impairments and mechanisms of interaction are less acknowledged at the levels of phonological and phonetic processes. Here, the author aims at bringing together empirical evidence in favor of an interaction between phonological and phonetic encoding and of the predominance of mixed patterns of impairment over pure phonetic impairment. Method: The author reviews empirical results from acoustic and psycholinguistic studies, both with healthy speakers and speakers with brain damage, favoring independent phonological and phonetic encoding and separable impairments as well as recent research pointing to an interaction between phonological and phonetic encoding processes and overlapping patterns of impairments. Conclusions: Acknowledging interaction between phonological and phonetic processing has clear consequences on the definition of patterns of impairment. In particular, phonetic errors have not necessarily a phonetic origin, and most patterns of impairment are bound to display both phonological and phonetic features.