A new inference mechanism using bi-level weights sum method for clinical computer assisted shock diagnosis algorithm is proposed in this paper. Shock is a very emergency physiological sign in clinical medicine. It predisposes to multi-organ failure. This inference method provides completely shock trend for clinician’s judgments. We use seven paths to infer the type of shock, including hypovolemic shock, cardiogenic shock, septic shock, anaphylactic shock, neurogenic shock, endocrine shock, and pulmonary shock. A knowledge base is composed of many possibility weights that are built by experienced medical expert, each path has further detail items and every item has respective weights for each shock type. Some items are then spilt into server, moderate and mild. In this study, nine patients’ data are collected and analyzed. The results provide order of shock type by bi-level weights sum method. The inference results computed by this system are coinciding with diagnosis by clinician and imply other potential shock type. These results are important for clinician because the results are not unique and which corresponds with shock physiological condition. We also distribute patients’ data to another six doctors for diagnostics, and for evaluating system performance. Results reveal it can provide sufficient complete information for clinician to ensure good diagnostics and treatment for patients. This system could be used either as a clinical decision support system or as educational software.