This work aimed at updating the estimates of genetic parameters for milk and protein yield in the Spanish Friesian population by making use of recently developed procedures that allow the utilization of massive data sets. This avoids the use of partial information when estimating variance components with respect to the information used in genetic evaluations. Data used in this study were 1,460,559 and 1,084,886 records from the January, 1998 national genetic evaluation for milk and protein yields. Pedigrees for the milk and protein analyses consisted of 703,326 and 589,385 animals, respectively. Variance components estimation was carried out under the same model used for the genetic evaluation, a repeatability animal model. Gibbs sampling within a Bayesian analysis was used to make inferences about the parameters of interest. The posterior mean for the heritability was 0.23 for milk yield and 0.22 for protein yield. The posterior mean for the repeatability for both traits was 0.46. Heritability estimates were slightly lower than previous estimates obtained in the same population with maximum likelihood procedures and also lower than the value currently used in the genetic evaluations. Additional research is needed to investigate the reasons of the relatively low heritability estimates obtained previous to propose its use in the national evaluations.