Tuberculosis is a chronic infectious disease caused by Mycobacterium tuberculosis. TB could cause serious health problems to the community if it remains undiagnosed or misdiagnosed and untreated. When regular TB is inadequately diagnosed and treated, which causes bacteria to develop resistance to the drugs used called MDR-TB and XDR-TB. For this reason, measures are needed to control the transmission of the disease. The best method to control TB transmission is to establish a diagnosis as early as possible, and particularly to ensure complete cure. Early diagnosis of TB is the most effective tool available to reduce transmission. But in middle and low income countries like Ethiopia face a severe lack of medical doctors and medical specialists for early diagnosis of PTB, lymph node and DR-TB. KBS have become increasingly popular in a wide variety of medical applications. A number of problem solving methodologies are available for the development of knowledge based system. CBR is one of the important approaches for the development of KBS, which emphasizes the role of prior experience during future problem solving. This study makes an attempt to design and develop prototype knowledge based system using CBR approach that can support domain experts’ decision in order to facilitate the diagnosis and treatment of TB patients. In order to achieve the objective of this study, knowledge is acquired using both structured and unstructured interview with domain experts followed by relevant document analysis. After the knowledge is acquired, the knowledge is modeled using the formalism of CommonKADS. Then, the modeled knowledge is represented using CBR techniques and finally implemented using jCOLIBRI tool. The performance of the prototype system is measured using statistical analysis and end user acceptance testing. Consequently, the prototype system performance measures a recall of 74% and precision 83% and 86% domain experts accepted the prototype system for diagnosis of TB. Thus, the prototype system achieves the objectives of the research and has a good performance. However, in order to apply the system for diagnosis and treatment of TB additional investigation is required like adding general explanation facilities by integrating with other AI methodologies