Fuel rod design verification is the evaluation process of fuel rod safety performance during operation in reactor, in which the uncertainty of input parameters has important effect on evaluation results. To study the uncertainty systematically, fuel rod performance analysis software has been coupled with Dakota software to carry out fuel rod design verification, the results of nonparametric Monte Carlo and Latin Hypercube Sampling have been compared with those of traditional method. It turns out that the fuel rod inner pressure criterion is vulnerable to be under challenge for the reason of input uncertainty under consideration by traditional method. The defects can be made up by statistical nonparametric sampling, by which a larger safety margin is obtained, and a theoretical basis for fuel rod safety and economic performance enhancement is provided. Meanwhile, the temperature calculation result obtained by two sampling methods can be more referential compared with traditional method. For the cladding corrosion and strain criterion, the results of sampling methods and traditional method show no significant difference, for the reason that the uncertain input parameters are selected suitably. In conclusion, the statistical method based on nonparametric sampling can be more practically significant for safety performance evaluation of fuel rod in operation.