BACKGROUND: Medication self-management in stroke patients is important to prevent further progression of disease and incidence of side effects. The purpose of this study was to create a formula for predicting medication self-management introduction in stroke patients using functional independence measure items and patient data, including medication-related information. METHODS: This was a retrospective analysis of 104 patients (cerebral infarction, cerebral hemorrhage, subarachnoid hemorrhage) discharged from the Kaifukuki rehabilitation ward at Showa University Fujigaoka Rehabilitation Hospital from January to December 2012. Multivariate analysis was performed to develop a formula for predicting achievement of medication self-management. RESULTS: Of the 104 patients, 39 (37.5%) achieved medication self-management. In the logistic regression analysis, number of drugs, age, walk/wheelchair mobility FIM, and memory FIM were extracted as significant factors independently contributing to achievement of medication self-management (p < 0.05). The prediction formula was [4.404 - 0.229 × number of drugs at admission + 0.470 × walk/wheelchair mobility FIM at admission + 0.416 × memory FIM at admission - 0.112 × age]. CONCLUSIONS: In the future, this formula may be used as an index to predict success of medication self-management in stroke patients.