Fusarium Head Blight (FHB) is a devastating disease that can cause severe loss of grain yield and quality of durum wheat in the northern Great Plain of the US. FHB resistance in wheat is a complex trait controlled by many genes. Recurrent selection is an effective way to increase frequencies of favorable resistant alleles and to develop improved germplasm. In this study, four cycles of recurrent phenotypic selection were conducted for reducing FHB severity from 2019 to 2022 in a durum wheat population derived from intercrossing of 15 elite cultivars and breeding lines. The FHB severity was reduced 34.5% from Cycle 0 to Cycle 3 population. Significant negative correlations were found between FHB severity and both plant height and days to flowering in Cycle 0, Cycle 3 populations. Genomic selection can speed up selection and increase genetic gain in terms of time and cost. A total of 284 S0 parents from the Cycle 2 and Cycle 3 populations were genotyped using 90K SNP array and obtained 2,706 SNP markers. Using ridge regression best linear unbiased prediction (rrBLUP) model, the prediction accuracy for FHB severity was 0.53 with cross-validation. Our results indicate that recurrent phenotypic selection can improve FHB resistance in durum wheat. Implementing genomic-assisted selection in the recurrent selection is practical to accelerate genetic improvement.