This article aims at determining the statistical validity of various constraints which are held to influence the element ordering of coordinate lexical items in English. Twelve constraints mentioned in the literature are tested on 562 binomials, compounds and blends. Nine of them are found to be statistically significant: experiential closeness, temporal iconicity, syllable number, vowel length, initial consonant obstruency, final consonant obstruency, stress alternation and lexical frequency for binomials; temporal iconicity and syllable number for compounds; syllable number and initial consonant complexity for blends.