Kern, Elisandra Lurdes [UNESP]Cobuci, Jaime AraujoNeto, José BracciniDos Santos Daltro, Darlene2019-10-062019-10-062019-01-01Animal Production Science, v. 59, n. 8, p. 1546-1552, 2019.1836-57871836-0939http://hdl.handle.net/11449/188376The objective of the present work was to evaluate the effect of somatic cell score on the longevity of Holstein cows raised all over Brazil, using a piecewise Weibull proportional-hazard model. The following two longevity traits were defined: True longevity, number of days from first calving to culling and functional longevity, approximated by correcting true longevity to within-herd-year production. Records on productive life of 131 330 cows were used. The model included the time-independent effect of age at first calving. The other effects were time-dependent, and included the following: Region by year of calving, variation in herd-size class, milk-production class by year of calving within herd, within-herd milk-production class by number of lactations, within-herd fat content, within-herd protein content, and somatic cell score (SCS). The overall mean of the somatic cell counts (SCC) was 322 000 cells/mL. The highest SCC mean was found between 130 to 290 days of lactation. The SCC mean decreased over the years. Cows from Region 5 (Rio Grande do Sul) showed higher SCC means. The risk of culling was slightly higher for functional longevity than for true longevity. The impact of longevity was high in cows from first to fourth lactation with a high SCS, with the risk of culling varying from 0.90 (true longevity: Second lactation and Class 2) to 1.2 (functional longevity: Fourth lactation and Class 5). Cows at the fifth lactation with a lower SCS had a higher risk of culling (1.4). Including the effect of SCS class by stage of lactation in the models was not beneficial. The decrease in SCS, especially from the first to fourth lactation, can be used for indirect selection to improve the longevity of Holstein cows in Brazil.1546-1552engDairy cattleFunctional longevityRisk of cullingSurvival analysisArtigo10.1071/AN18069Acesso aberto2-s2.0-85056740093