Three experiments were conducted to test the phonological recoding hypothesis in visual word recognition. Most studies on this issue have been conducted using mono-syllabic words, eventually constructing various models of phonological processing. Yet in many languages including English, the majority of words are multi-syllabic words. English includes words incorporating a silent letter in their letter strings (e.g., champa"g"ne). Such words provide an opportunity for investigating the role of phonological information in multi-syllabic words by comparing them to words that do not have the silent letter in the corresponding position (e.g., passen"g"er). The performance focus is on the effects of removing letters from words with a silent letter and from words with a non-silent letter. Three representative lexical tasks--naming, semantic categorization, lexical decision--were conducted in the present study. Stimuli that excluded a silent letter (e.g., champa_ne) were processed faster than those that excluded a sounding letter (e.g., passen_er) in the naming (Experiment 1), the semantic categorization (Experiment 2), and the lexical decision task (Experiment 3). The convergent evidence from these three experiments provides seminal proof of phonological recoding in multi-syllabic word recognition.