Blood and milk samples were collected in 1992/93 from 2-year-old heifers in the Livestock Improvement Corporation's Sire Proving Scheme, for analysis of carotenoid concentration (CC). The trial comprised heifers in 127 spring-calving North Island dairy herds, where blood samples were taken from a total of 2744 heifers (Holstein-Friesians (F), Jerseys (J), and their crosses) in early/mid lactation (spring), and from a sub-group of the same animals in autumn. These heifers were the daughters of 157 young F and J sires, and 15 older sires. Whole-lactation yields of milk, fat, and protein, and fat% and protein% were also obtained. Sire-model restricted maximum likelihood (REML) analyses were carried out to estimate genetic and phenotypic parameters. Purebred F and J breed means for plasma CC in spring were 14.1 and 17.9 mug ml(-1) (P < 0.001), respectively, and for milk CC 5.24 and 7.50 mug g(-1) fat (P < 0.001), and corresponding heritabilities for the two traits were 0.46 +/- 0.08 and 0.11 +/- 0.10. The repeatability of plasma CC across seasons was 0.64 +/- 0.02. The genetic correlation between spring plasma CC and milk CC g(-1) fat was 0.66 +/- 0.22, whereas genetic correlation estimates for spring plasma and milk CC g(-1) fat with other traits were: with milk volume -0.01 +/- 0.16 and 0.41 +/- 0.35, with fat yield 0.06 +/- 0.22 and 0.29 +/- 0.54, and with fat% 0.03 +/- 0.13 and -0.26 +/- 0.29. It was concluded that plasma CC was a repeatable and heritable trait, whereas milk CC g(-1) fat was lowly inherited; plasma CC was essentially uncorrelated with the standard milk production traits; if bulls were selected for lower plasma CC, there would be reduced CC in both the plasma and milk of their daughters.