<p>Rheumatoid arthritis (RA) is a debilitating autoimmune disease affecting over 10&#xa0;million people globally, imposing a heavy disease burden and creating urgent demand for online health information among patients. TikTok and Bilibili have become significant information carriers due to their dissemination characteristics, but a lack of strict regulation raises concerns about content quality. Research on the quality of RA-related short videos on Chinese platforms remains unexplored. This study aims to evaluate the quality of relevant content on both platforms to guide patients in accessing evidence-based recommendations. This cross-sectional study used the Chinese search term “rheumatoid arthritis” to collect relevant short videos from TikTok and Bilibili. To minimize bias, newly created accounts collected the top 100 videos from each platform using default sorting in incognito mode. After excluding advertisements, irrelevant content, videos uploaded within the past week, and duplicates, 180 videos were ultimately included (94 from TikTok, 86 from Bilibili). Variables extracted included video source, engagement metrics (likes, comments, collections, shares), duration, and uploader type. Video quality was assessed using a modified DISCERN scale (evaluating clarity, relevance, etc., across 5 criteria, scored 0–5) and the Global Quality Score (GQS, 1–5 Likert scale) to evaluate video quality. Content completeness was assessed across six dimensions (epidemiology, etiology, symptoms, etc.) using a 3-point scale (0 = not addressed, 1 = partially explained, 2 = fully explained). Assessments were conducted by two uniformly trained medical professionals and one RA patient. All raters completed the evaluation according to the same standardized process. Data analysis employed descriptive statistics, Mann-Whitney U tests, Kruskal-Wallis H tests, Spearman correlation analysis, and multiple linear regression with 1000 bootstrap resampling, BCa 95% confidence intervals, and variance inflation factor (VIF) for multicollinearity testing using IBM SPSS Statistics 26.0 and GraphPad Prism 10.1 software. P &lt; 0.05 indicated statistically significant differences. TikTok had significantly higher user engagement than Bilibili, while Bilibili videos were significantly longer in duration. Professional individuals were the main uploaders of RA-related videos, and TikTok had a higher proportion of professional uploaders than Bilibili. In terms of content completeness, TikTok performed well in symptoms and treatment dimensions, and Bilibili was superior in etiology, symptoms and diagnosis dimensions. Both platforms were seriously deficient in prevention-related content. Bilibili had significantly higher video quality scores than TikTok. Multiple linear regression identified independent predictors of user likes after multicollinearity adjustment. Correlation analysis showed good consistency of inter-rater evaluation. This study first demonstrates that Bilibili provides significantly more complete and higher-quality RA-related short videos than TikTok. Although TikTok shows higher user engagement, this elevated engagement correlates with the wider spread of low-quality content, which may limit access to accurate medical information and impair evidence-based treatment decisions. Both platforms lack sufficient preventive content, weakening early intervention and long-term disease management. These discrepancies arise from platform-specific user groups, content formats, and algorithm mechanisms. To mitigate the adverse effects of misleading information on RA management and treatment adherence, we recommend algorithm optimization, professional creator certification, and targeted health education. Multistakeholder collaboration to promote evidence-based content and systematic programs to improve digital health literacy are also essential. This study offers robust empirical evidence for standardized RA health communication on short-video platforms, facilitating access to accurate information and supporting rational clinical decision-making for patients. </p>