The purpose of this study was to develop a method for automatic and stable determination of the optimal time range for fitting with a Patlak plot model in order to measure myocardial perfusion using coronary X-ray angiography images. A conventional two-compartment model is used to measure perfusion, and the slope of the Patlak plot is calculated to obtain a perfusion image. The model holds for only a few seconds while the contrast agent flows from artery to myocardium. Therefore, a specific time range should be determined for fitting with the model. To determine this time range, automation is needed for routine examinations. The optimal time range was determined to minimize the standard error between data points and their least-squares regression straight line in the Patlak plot. A total of 28 datasets were tested in seven porcine models. The new method successfully detected the time range when contrast agent flowed from artery to myocardium. The mean cross correlation in the linear regression analysis (R²) was 0.996 ± 0.004. The mean length of the optimal time range was 3.61 ± 1.29 frames (2.18 ± 1.40 s). This newly developed method can automatically determine the optimal time range for fitting with the model.