OBJECTIVES: Disability presents a major economic and humanistic burden among stroke survivors. To quantify disability in patients, activities of daily living (ADLs) measures have been used. The translation among measures, however, has not been developed, leaving a significant discontinuation of patient care between facilities that use different measures. The purpose of this study is to develop a conversion system for two most widely used ADL measures, the Motor component in Functional Independence Measure (M-FIM) and the Barthel Index (BI). METHODS: We use Kansas City Stroke Study data. We standardized the scales of both instruments to 0-1 and rotated them 45 degrees in order to get a consistent conversion scheme regardless of the direction of the conversion (from BI to M-FIM or from M-FIM to BI). We applied the censored heteroscedastic regression spline model using Monte Carlos Expectation and Conditional Maximization (MCECM) algorithm, since the data is censored in both boundaries (for M-FIM: 13-91, and BI:0-100) with ceiling and floor effects. RESULTS: Among 1676 records, 2.4% and 5.8% of M-FIM records, and 3.5% and 24% of BI presented the lowest and highest scores respectively. Based on our model, zero, 60, and 100 of BI scores are equivalent to 13 to 15, 59 to 62, and 88 to 91 of M-FIM. (Complete conversion table will be provided in the presentation.) The two measures were highly correlated (r=0.9479, p<0.0001), but the linearity assumption was improper, particularly, for the higher score range of the M-FIM and BI. CONCLUSIONS: For the continuum of patient care, the conversion among scales is an important issue. We developed a conversion scheme between two widely used ADL measures, BI and M-FIM, based on their total scores. Further research is planned to obtain a more robust and precise conversion model addressing item level information.