Cardiac resynchronization therapy (CRT) is one of the effective treatment techniques to prevent cardiac arrest due to abnormal QRS electrocardiogram waveforms. However, CRT does not show good results in all clinical cases. In this study, we investigate a simulation based screening of CRT and build up a hypothetical treatment simulation model to predict the function and condition of the heart in post CRT condition. For this purpose, we first constructed patient-specific three-dimensional heart models with left bundle branch block. Then, the newly proposed method of CRT simulation incorporating CT and MRI images of patients was proposed. The present ventricular model consists of cardiac electric excitation and contraction model based on Hermite mesh system that was reconstructed from CT images. To calculate the electrophysiological stimulation of myocardial cells, mono-domain method with cell model was used. The present simulation model established a patient-specific model by matching measured data of systolic, diastolic blood pressure, ejection fraction (EF) of a patient with computed one. Here, we used MRI images to implement a realistic mechanical contraction in the model. Then, we applied this method to 7 CRT patients and compared the results with clinically measured ones. Computed results showed that the present CRT method can discriminate responder and non-responder of CRT and moreover identify optimal positions of ventricular pacing leads.