The objective of this study was to estimate genetic parameters for longevity and assess the suitability of using these selection criteria to improve the genetic merit of the beef cattle population of the Czech Republic. The performance record database, which contains records of 363,000 beef cattle animals of 19 breeds and their crosses, was used. The populations of Charolais and Aberdeen Angus were large enough that the genetic parameter estimations and all analyses were done for these breeds separately. Two similar approaches of longevity definition based on probabilities were considered as follows: productive longevity (PL), which is the number of calvings at target ages of 78, 90, 150, and 160 mo, and longevity (L), which is based on the probabilities of cow reappearance in the next parity. A multibreed single-trait animal model for L and a multitrait animal model for combinations of 78/150 and 90/160 mo for PL were used. Specific combinations of months were established based on the analysis and represented the critical culling rates in the studied population. The high genetic correlations (0.88-0.95) of the combination 90/160 suggested that the PL at 160 mo of age can be predicted on the basis of the value at 90 mo, which will make earlier selection possible. Combination 78/150 is less efficient in the Czech population of beef cattle due to the lower correlations (0.79-0.93) between traits. The estimated heritabilities were low for both traits (below 0.14), but the additive genetic variance was sufficient for identifying animals with high genetic merit.