Santiago, MacKay, Palma, and Rho (2000) report two picture naming experiments examining the role of syllable onset complexity and number of syllables in spoken word production. Experiment 1 showed that naming latencies are longer for words with two syllables (e.g., demon) than one syllable (e.g., duck), and longer for words beginning with a consonant cluster (e.g., drill) than a single consonant (e.g., duck). Experiment 2 replicated these ndings and showed that the complexity of the syllable nucleus and coda has no effect. These results are taken to support MacKay's (1987) Node Structure theory and to refute models such as WEAVER ‡ ‡ (Roelofs, 1997a) that predict effects of word length but not of onset complexity and number of syllables per se. In this comment, I show that a re-analysis of the data of Santiago et al. that takes word length into account leads to the opposite conclusion. The observed effects of onset complexity and number of syllables appear to be length effects, supporting WEAVER ‡ ‡ and contradicting the Node Structure theory.Theories of spoken word production differ in their claims about the role of syllables and their internal structure. Santiago, MacKay, Palma, and Rho (2000) contrasted the predictions of two types of theory. According to MacKay's (1987) Node Structure theory, henceforth NST, syllables and their internal structure are planned via a tree traversal process that operates in a top-down and left-to-right fashion. In particular, the initial syllable of a word is planned before its non-initial syllables, and syllable Requests for reprints should be addressed to Ardi Roelofs,