We examined the effect of two methods of assessment--multiple-choice sentence-picture matching and an animated sentence-verification task--on typically developing children's understanding of relative clauses. A sample of children between the ages of 3 years 6 months and 4 years 11 months took part in the study (N = 103). Results indicated that (a) participants performed better on the sentence-verification than on the multiple-choice task independently of age, (b) each testing method revealed a different hierarchy of constructions, and (c) the impact of testing method on participants' performance was greater for some constructions than others. Our results suggest that young children can understand complex sentences when they are presented in a manner that better reflects how people process language in natural discourse. These results have implications for the study of language comprehension in suggesting that results from multiple-choice tasks may not generalize to other methods.