Consumers often rate online product reviews in terms of helpfulness. To explore the linguistic features that may contribute to helpfulness ratings, a linguistic inquiry and word count analysis compared 377 helpful and unhelpful Amazon.com product reviews, showing that helpful and unhelpful reviews differed across 23 of 67 linguistic categories. Results also suggested a consumer preference for messages that contain objective characteristics and high-quality arguments. A follow-up study tested hypothesized relationships among helpfulness ratings and review relevance, descriptiveness, and evaluation discrepancy. The data showed that descriptive reviews were rated as helpful. Implications, limitations, and directions for future research are discussed.