Seventy-two sentences presented to ten preschool children for repetition were designed so that three sentence construction factors varied independently. The factors were (1) length in number of words, (2) complexity of personal pronouns and main verbs as scaled by Lee (1974), and (3) word familiarity, defined as common vocabulary or the substitution of a nonsense word in place of a typical noun or verb in the model sentence. Three methods were employed for scoring the children's responses: (1) number of retained words, (2) Developmental Sentence Scoring (Lee, 1974), and (3) Stephens's Categories (Stephens, 1974). Eighteen sentences were re-presented for the assessment of reliability. The results of multiple regression analyses indicated that length was the important contributing factor in the children's responses to the model sentences and that Stephens's Category Scale of response scoring was the most sensitive method for detecting the influence of the three sentence factors on the children's responses.