This study examined the effects of lexical frequency on children's production of accurate primary stress in words derived with nonneutral English suffixes. Forty-four third-grade children participated in an elicited derived word task in which they produced high-frequency, low-frequency, and nonsense-derived words with stress-changing suffixes (i.e., -tion, -ic, -ity). Derived word frequency ... [Show full abstract] affected stress production accuracy; however, the individual suffix also played an important role in stress placement, with -tion productions more accurate than either -ic or -ity productions. For the real words, derived word frequency relative to stem frequency was related to performance. Stress was less accurate on derived words that were much lower in frequency than their stems (e.g., tranquil/tranquility) and more accurate on derived words that approximated or exceeded their stem frequency (e.g., motivate/motivation). In addition to derived word and stem frequency, results are discussed with reference to several phonological characteristics that may also influence stress production accuracy.