Certain structures are particularly challenging for children. Explanations of such challenges reference both grammatical development and processing capacities. This study concerns production-specific considerations. Sixteen adults and 72 children from ages 3;01 to 8;11 participated in an experiment designed to elicit imitation of one-, two-, and three-clause structures in active and passive voice. The complex structures included complements and relative clauses. The experiment yielded 1,884 utterances that were coded for lexical, grammatical, and fluency factors. Articulation rate was calculated for the 790 fully fluent utterances. Age, utterance length, and structure affected rate. Increases in rate as speakers age are generally taken to reflect gains in proficiency as the complex skills of language production are practiced. The length effect is tentatively explained in terms of a strategy that adjusts the time to articulate different structures. Central analyses found passives articulated faster than actives, and subject (passive) relative clauses articulated faster than object (active) relative clauses. These structural effects--the rate increase in passives and rate decrease in object relatives--support a planning-based account that hinges on filler-gap dependencies.