Word knowledge has long been considered as one of the most important predictors of reading comprehension, academic achievement, and social development. However, it was relatively narrowly conceptualized and measured as either the number of words individuals know the general meaning of (i.e., breadth) or the simple association between two words (i.e., depth). The problem with such a one-sided view is that the existing measures of word knowledge is limited in revealing the quality of word knowledge, which is characterized by both depth and breadth. In this investigation, by comparison, quality of word knowledge was conceptualized as individuals’ fine-grained understanding of the word meanings as they systematically identified semantic similarities and differences among a group of words. It is believed that through deliberate comparisons among word sets, individual’s understanding of the intricacy, subtlety in the meaning of words can be revealed. Therefore, this study offered a new approach in assessing the word knowledge quality informed by a theoretical model of relational reasoning and its four resulting forms (Alexander & DRLRL, 2012). A novel measure of word knowledge quality, Relational Reasoning with Words (R2W2) was developed and validated in this study. Moreover, the unique contributions of relational reasoning ability and word knowledge to college students’ performance on R2W2 were also analyzed. With a sample that involved 338 participants from four US universities, the study found that R2W2 was a reliable and valid measure for word knowledge quality with sound psychometric properties on the item level. In addition, word knowledge was found to contribute to college students’ performance on R2W2 more than relational reasoning ability. Implications for future research and practice are also presented and discussed.