Breeding values (BVs) for four plant (bloom date, fruit development period, fruit density, and blind node propensity) and five fruit (weight, blush, shape, soluble solids, and titratable acidity) traits of 28 peach [Prunus persica (L.) Batsch (Peach Group)] genotypes used as parents in the Texas A&M University peach breeding program were predicted using best linear unbiased prediction (BLUP). Data from seedlings of 108 families developed from 42 peach parents were analyzed by using a mixed linear model, with years treated as fixed and additive genotypes as random factors. The precision of the predictions was high for most parental genotypes, as indicated by the correlations (rTI) between predicted and true BVs and the standard error of the predictions (SEP). In most cases, the higher the number of progeny, the better the agreement between predicted and true BVs for that parent. Parents with observations from more than 30 seedlings had a rTI ≥ 0.90 and smaller SEPs. For all traits analyzed, the lowest precision (low rTI and high SEP) was observed for `Flordaking', whose predicted BVs was based only on pedigree information.