Grey relational analysis (GRA), an approach totally different from traditional statistical analysis used to calculate the relational grade between two sequence data, was applied to classify dried roselle (Hibiscus sabdariffa L.). In this study, 49 groups of roselle calyces representing five treatments – (1) frozen at −20C; (2) cold‐air‐dried at 25C, or (3) hot‐air‐dried at 50C; (4) 75C; or (5) 85C – were investigated. Five calyces from each group of treated roselle calyces were selected at random as reference samples and were built into the database. The rest of the treated roselle calyces were selected as test calyces to validate the feasibility of the developed Grey identification system. A GRA calculation procedure is demonstrated step‐by‐step and results are compared with the results of principal component analysis (PCA). PCA exhibited no distinguishable cluster pattern on the PCA plot of roselle calyces hot‐air‐dried at 50C or 75C. The GRA grades identified and classified all undried and dried roselle calyces correctly.