In a hybrid plant breeding program, mixed-model analysis is a powerful tool for the identification of superior hybrids and inbreds. The genetic determinants of hybrid performance can be partitioned into the additive and non-additive genetic components. The additive component is a measure of the breeding value or general combining ability (GCA) of the inbreds while the non-additive genetic component reflects the specific combining ability (SCA) of the hybrids. Estimates for the GCA values of the inbreds and SCA values of the hybrids can be obtained using the Best Linear Unbiased Prediction (BLUP) method. In interpreting these values, it is of critical importance to have some measure for assessing how confident we can be in each predicted value. This measure should take into account the amount of information for each individual as well as its genetic relationships with the other individuals. An approach developed for estimation of breeding values in animal breeding is based on the Prediction Error Variance (PEV) where the PEV = var(a - â) where a is a random effect and â is the predicted value for that random effect. By regarding the PEV as the fraction of the variance for a given genetic component that is not accounted for by the BLUP values, the PEV can be equated with (1 - r²)σ² where r² is the squared correlation between the true and predicted values and σ² is the variance of the genetic component. In animal breeding, r² is referred to as the reliability of a predicted value. Reliability measures obtained from a maize breeding program are presented.