The accuracy of diagnosis coding has a significant impact on the BPJS claims process. This study aims to investigate the relationship between the accuracy of stroke diagnosis codes and the approval of claims by the Social Security Administration for inpatient care. This type of research is analytical research with a cross-sectional approach. The statistical test used in this study is Fisher's Exact Test. The population was 506 medical records, and a sample of 83 medical records was selected using a simple random sampling technique. The research instruments were observation and interview guidelines, checklists, ICD-10, and SPSS version 26. Data processing included data compilation, editing, coding, classification, tabulation, and data presentation. The results of the analysis of existing coding procedures were documented in Standard Operating Procedures (SOP). The accuracy of stroke diagnosis codes was 70 (84%) of medical records, while the inaccuracy of stroke diagnosis codes was 13 (16%) of medical records. The percentage of BPJS claims approval was 52 (63%) medical records, while those who did not approved were 31 (37%) medical records. The Fisher Exact Test calculation results showed a p-value <0.014 significance level (0.014 <0.05) in SPSS. Conclusion: There is a relationship between stroke diagnosis code accuracy and BPJS claim approval. It is recommended to revise the SOP in point 3, requiring coding staff to reconfirm with physicians if they encounter any issues in the coding process and the management procedures that support the diagnosis. The head of medical records should recommend that coding staff attend coding training.