In two experiments, we investigated whether onsets and rimes have a role in the processing of written English. In both experiments, participants detected letter targets (e.g., t) in nonwords like vult faster than in nonwords like vust. This finding is consistent with Selkirk's (1982) view that sonorants (e.g., the III of vult) cohere with preceding short vowels and are part of the vowel nucleus. In contrast, the ItI of oust is part of the syllable's coda st and so is harder to isolate. Experiment 2 demonstrated that the time required for one to detect single-member codas following vowel digraphs (e.g., the t in veet) was similar to the time to detect the same target letter following a postvocalic sonorant (e.g., the t in vult). No evidence was found for onsets. The results provide support for a phonological organization among letters of printed rimes.A major goal of psycholinguistics is to understand the units involved in the processing of spoken and written English. In speech processing, one level of analysis of analysis is the word. Spoken words also are divisible into syllables and phonemes. Compatibly, English orthography has a lexical level (represented by the spaces between words) and a graphemic level where a letter or a group of letters represents a single sound. Whole words map to complete phonological representations (e.g., aisle-oteul).Graphemes map to individual phonemes (e.g., ck~/k/). Grapheme-phoneme correspondences may exist at an intermediate level, larger than individual graphemes but smaller than words (e.g., break-eiott + leik/). This hypothesis is investigated in the present study.