The objectives of this study were to (a) evaluate changes in speech intelligibility in a group of 110 adolescent users of cochlear implants who were first assessed in elementary school (CI-E) and later in high school (CI-HS) and (b) examine factors influencing speech intelligibility performance at the CI-E and CI-HS sessions. Participants were 110 adolescents who participated in an earlier study examining 181 young elementary school-aged children. Primary outcome measures included speech intelligibility under quiet and multispeaker background conditions and consonants correct produced in the sentences. Multiple linear regressions were used to evaluate how participant, family, and performance measures influenced their speech production during adolescence. Performance measures included estimates of speech perception, working memory, sign enhancement, and duration of seven-syllable sentences. Participant and family measures included duration of deafness, performance intelligence quotients, gender, family size, and socioeconomic status. Principal component analyses were used to construct common variables across highly intercorrelated measures. Three sets of multiple linear regressions evaluated the contributions of the variables to the variance associated with adolescent speech intelligibility. Speech intelligibility and consonants correct significantly increased nearly 22% between the two test sessions. Speech intelligibility significantly decreased by approximately 20% in the multispeaker babble condition relative to the quiet condition. Duration of seven-syllable sentences significantly decreased during the two test sessions. Data revealed that 65.8% of the variance in adolescent speech intelligibility was predicted from participant, family, and performance measures observed in elementary school. Forty-nine percent of the variance at adolescence was accounted for by the participant, family, and performance measures observed during the high school test session. Evaluation of variance including participant and family measures at both time periods, in conjunction with the adolescent performance measures, accounted for 49% of the variance in adolescence performance. After contributions from participant and family variables at the elementary and adolescent test sessions were removed, 21% of the variance in adolescent speech intelligibility was due to the performance measures at adolescence. Independent predictors of performance at adolescence included negative effects of sign enhancement and duration of seven-syllable sentences. Substantial improvements were made in consonant accuracy, sentence duration, and speech intelligibility between elementary and high school test sessions. Reductions in speech intelligibility performance suggest that allophonic variations, distortions, or use of speech sounds in a nonambient language may contribute to the reductions observed in multispeaker background conditions. Although a significant amount of variance in adolescent performance is accounted for by participant and family characteristics, elementary school speech production and an early reliance on speaking and listening independently account for variance in adolescence speech intelligibility. Over and beyond all the contributions made by participant, family and performance measures, greater reliance on oral communication, and shorter sentence durations independently account for variance at adolescence.