Abstract: During the normal operation of pressurized water reactor, fission products activity in the primary coolant will increase when failure of fuel rods happens. The prediction of the status of fuel failure has been paid great attention from industry for the long term with respect to radiological protection, radioactive consequence and the economy. The iodine isotope activity ratio 131I/133I is one of the most important indicators to evaluate fuel reliability in nuclear power industry. In this paper, the production and release of fission products in fuel rods and the primary coolant were simulated by the kinetic model and the typical 131I/133I in the primary coolant were theoretically estimated at the equilibrium conditions for the intact fuels, fuel failure with small defect and large defects, respectively. The radiochemical data in the primary loops were gathered and compiled from thirty six cycles in operating CPR1000 PWR units. The statistical results of these operation data show that the higher volume activities of radio-iodine in the primary loops may result from the dissemination of actinides due to secondary hydriding from the previous cycles and the 131I/133I can be used to well identify the fuel reliability in comparison of the radio-iodine activities. In the cycles with fuel failure, the radio-iodine activities and 131I/133I according to the operation data distribute much wider than the expected because the actual status of fuel failure are more complicated during operation than expected, e.g. changes of fuel failure size, the discrepancy of the location of fuel failure and the relative power of the defective fuel rods. In some cases, it is difficult to distinguish the fuel failure by radio-iodine owing to its low release rate from defective fuel rods. In addition, the size of fuel failure in the specific cycles can be illustrated by the significant change of 131I/133I. It shows that the predictions by the model and the statistical results in operating CPR1000 PWR units are qualitatively in agreement for both intact fuel and fuel failure. It also indicates that the conventional threshold 131I/133I≥0.1 for fuel failure may make the misjudgment due to the overlap of distributions and 131I/133I≥0.15 can distinguish 97% operation date for the fuel failure and 98% operation date for intact fuel rods. The method used and the results in the paper can help to make better understanding of the release of iodine for fuel failure and identity the defective fuel rods during unit operation.