In the paper, a formulation for predicting the ultimate strength of a stiffened plate is proposed by incorporating Guedes Soares’s formula, which gives the best prediction for unstiffened plates according to the calibration carried out recently by the authors, into Faulkner’s method (Faulkner et al., 1973). The proposed algorithm is then calibrated by using a considerable amount of experimental and numerical data. It is observed that: (a) The proposed method shows better prediction than Faulkner’s original method if only the experimental data (63 samples) are included in the calibration, the bias and COV of the model uncertainty of the proposed method are 0.992 and 0.099, respectively, while they are 1.039 and 0.143 for Faulkner’s original method, and the skewness of the proposed method is small (only −0.105 slope, which is defined as the slope of the regressed straight line on the plot of model uncertainty against predicted value). (b) On the whole, including experimental and numerical data, the results of the proposed method demonstrate more or less the same accuracy as that of the original Faulkner method with better bias and skewness, but slightly larger scatters than the original Faulkner method. In addition, the reliability analyses of stiffened plates are carried out by using advanced first-order second-moment method (AFOSM), the second-order reliability method (SORM), and Monte Carlo simulation to investigate the accuracy of the first and second-order methods. It is found that the difference between the two methods is so small that the values obtained from AFOSM are acceptable in practice, considering the nominal nature of the reliability index.