We previously reported reliability and validity of our newly developed comorbidity scale (CS) for stroke outcome research based on a retrospective sample. The objectives of this study were to cross-validate the comorbidity scale in a new prospective sample and to investigate longitudinal changes of the comorbidity scale during hospitalization. In a prospective sample of 175 stroke patients admitted to five nonacute rehabilitation hospitals in Japan, we analyzed the frequency and grading of comorbidities and compared the comorbidity scale with demographic data, impairment as assessed with the Stroke Impairment Assessment Set (SIAS), and disability as measured with the Functional Independence Measure (FIMSM). The results were compared with our previous retrospective study. We also studied longitudinal changes by measuring the comorbidity scale on admission, 2 wk later, and at discharge of 67 patients. As a result, the comorbidity scaling was significantly lower in the prospective sample, and it increased at the second measurement and then plateaued. Among the comorbidities, hypertension ranked first, followed by shoulder pain, and diabetes mellitus. Similar to our previous retrospective study, the comorbidity scale correlated positively with age and length of stay and correlated negatively with the SIAS motor item scores and the FIM scores. In conclusion, the present study suggested concurrent and predictive validity of the comorbidity scale in a prospective sample as well and clarified the comorbidity characteristics of stroke inpatients.