Genomic prediction is most useful in early generations, but the family structure in these generations can reduce selection efficacy. To test genomic prediction strategies at this stage, a population representing an early generation breeding program stage (450 genotypes) was planted in two locations in November of 2023 and harvested at the beginning of June in Baton Rouge and Winnsboro, Louisiana. During the field season, heading date was collected in both locations, and plant height was collected in Winnsboro. Visual FHB ratings were taken after the appearance of the disease. After harvest, Fusarium damaged kernel (FDK) rates were taken on harvested grain, and ground samples were assayed for Deoxinivalenol (DON) content. Multi-trait models utilizing correlations between traits were used to generate predictions between and within families.