||The development of limit-state design codes for large-diameter bored piles requires a careful calibration of design methods in use. Both capacity and serviceability requirements must be satisfied in limit state design. Due to various source of uncertainty in subsurface conditions, construction effects, errors in obtaining soil or rock properties, and imperfection of prediction models, the estimated capacity and settlement values from various design methods vary widely and the corresponding reliability levels are not clearly known. This study aims to evaluate the performance of design methods for predicting the capacity and settlement of large-diameter bored piles. A database of load tests on large-diameter bored piles in Hong Kong is collected and used as the basis for performance evaluation. To facilitate the design for the ultimate limit states, six pile-capacity prediction methods for bored piles founded in soils and four methods for piles socketed in rocks are evaluated. The capability of these design methods to estimate individual resistance components and the total capacity is separately evaluated. Construction effects, particularly the use of permanent liners, pile-shaft grouting, and time delay between the end of pile excavation and concreting, are explicitly considered. Several empirical correlations are established based on the results of pile-load tests in the database. The performance of these empirical correlations is also evaluated. To facilitate design for the serviceability limit states, which is an important area under research, four pile settlement analysis methods for bored piles founded in soils and three methods for bored piles socketed in rocks are evaluated. The calibration exercises are conducted at the design load level. The results show that the coefficients of variation for the analysis methods for rock-socketed piles are smaller than 0.30, which indicates higher accuracy of prediction compared with the ability to estimate the pile capacity. Finally, resistance factors or partial factors for the design methods for pile capacity prediction and pile settlement prediction are developed using reliability principles based on the results of the above extensive model calibration exercise.