Identifying helpful drug reviews could significantly assist patients in their medication decision-making. Despite that review helpfulness has been extensively explored in the prior literature, the findings might not be applicable to drug reviews due to the considerable medical-specific characteristics. In this study, we leveraged signaling theory and developed a theoretical framework to reveal how different information signals regarding patients’ health conditions influence perceived review helpfulness. We harvested a large drug review dataset covering 9,908 drugs and corresponding 147,169 reviews from WebMD and adopted a deep learning technique to extract sentence-level medical signals directly from drug reviews with promising performance. Our empirical analysis suggests that information signals related to patients’ health conditions and medical experiences have significant positive impacts on reviews’ perceived helpfulness. In addition, their impacts largely depend on the drug type, review volume, overall rating, and how long the reviewer has taken the drug, revealing under which conditions the effects of patients’ health signals on review helpfulness are likely to be weakened or strengthened. This study provides both theoretical and methodological contributions to the research on the helpfulness of reviews, especially in the online medical review context, and provides practical implications for various stakeholders regarding medication decision-making.