Letter migrations between words in reading aloud (e.g., reading “cane love” as “lane love” or “lane cove”) are known to result from a deficit in the visual-orthographic analysis and characterize attentional dyslexia. In spontaneous speech, individuals with impairment in the phonological output buffer may show migrations of phonemes between words. The purpose of this study was to examine whether migrations between words in reading aloud can also result from a deficit in the phonological output buffer, to explore the characteristics of migrations resulting from orthographic input and from phonological output deficits, and to examine methods to distinguish these two sources. Using tasks of reading aloud of lists of 92–182 word pairs, we identified 18 adults and adolescents with developmental dyslexia who made between-word letter migrations in reading aloud, significantly more than age-matched controls (372 adults, 26 7th-graders and 44 4th–5th-graders). To distinguish between the orthographic input and phonological output sources for these migrations, we administered a test assessing orthographic input without spoken output (written semantic decision on 140 migratable word pairs) and a repetition test of 36 auditorily presented migratable word pairs, assessing spoken output without orthographic input (as well as nonword repetition and 3 span tests). These tests indicated that the migrations in reading aloud of 10 of the participants with dyslexia resulted from an orthographic input deficit—they made migrations not only in reading aloud but also in written word pair comprehension, but not in word pair repetition. For the other 8 participants, the migrations resulted from a phonological output deficit: they made migrations in reading aloud and in word pair repetition, but not in comprehension, and had limited spans and made errors in nonword repetition. We identified several differences between the two types of between-word errors: first, the individuals with attentional dyslexia made omissions of a letter that appeared in the same position in the two words, but the phonological output buffer group did not make such omissions. Second, the groups differed in the origin of migration: orthographic input migrations involve letters that are orthographically adjacent, whereas phonological output migrations involve phonologically adjacent phonemes: phonemes that have just been spoken or that are prepared together in the phonological buffer for production. Migrations from the line below and from 2 lines above the target occurred only in the orthographic input group. This study thus indicates that between-word migrations in reading aloud can result not only from attentional dyslexia, but also from a phonological output buffer deficit, and offers ways to distinguish between the two.