Response times have often been used as ancillary information to improve parameter estimation. Under the dual processing theory, assuming reading comprehension requires an automatic process, a fast, correct response is an indicator of effective automatic processing. A skilled, automatic comprehender should be high in response accuracy and low in response times. Following this argument, several questions were addressed in this study. First, individuals with higher ability endorsed a correct response more quickly in the reading comprehension assessment, suggesting "correct" response times provide useful information in discriminating individuals of different abilities. Second, in terms of predicting an external criterion of reading proficiency, the increment of predictive validity of categorized "correct" response times was larger than the incremental validity of continuous response times based on both correct and incorrect response times. An index reflecting both accuracy and response time yielded higher incremental validity than indices reflecting response time only. Results support response time as manifesting a dimension of interest in its own right, over its inclusion as an ancillary dimension in a multidimensional model.