Background: Several retrospective studies have previously demonstrated that a combination of adjuvant systemic chemotherapy and hepatic arterial infusion (HAI) could benefit patients following colorectal cancer liver metastases (CRLM) resection. This prospective clinical study aimed to determine whether adding HAI to adjuvant systemic chemotherapy could reduce the risk of recurrence following CRLM resection. Methods: The HARVEST study is an investigator-initiated, prospective, randomized controlled trial investigating the efficacy and safety of adjuvant intravenous chemotherapy with or without HAI floxuridine (FUDR) in CRLM patients that underwent liver metastasectomy. Patients in the systemic chemotherapy plus HAIC arm (HAI group) received systemic FOLFOX (q2w) plus HAI (FUDR, d1-14, q4w) for up to 6 months, while the systemic chemotherapy group without HAI (non-HAI group) received intravenous FOLFOX only. The primary study endpoint is the relapse-free survival in the modified intention-to-treat (mITT) population. Blood samples at different time points were also collected and circulating tumor DNA (ctDNA) was tested for NPY and SEPT9 methylation. Results: The study was prematurely terminated due to FUDR production halt in China. Ninety-two patients were randomized and seventy-seven patients (38 in the HAI group and 39 in the non-HAI group) were eventually included in our mITT analysis. After a median follow-up of 35.8 months, there were 22 (57.9%) and 25 (64.1%) recurrences in the HAI and non-HAI groups, respectively. The median relapse-free survival was 20.0 months in the HAI group and 11.7 months in the non-HAI group (p = 0.14; HR 0.65; 95% confidence interval [CI] 0.37 to 1.16). No significant difference was found in terms of overall survival between the two groups (p = 0.461). Our subgroup analysis revealed that patients with multiple liver metastases (p < 0.01) and RAS/BRAF mutation (p < 0.01) could benefit from adjuvant HAI treatment. Based on ctDNA status, patients with positive postoperative ctDNA methylation benefited from adjuvant HAI treatment (p < 0.01), while those with negative postoperative ctDNA methylation status (p = 0.95) did not. Chemotherapy-related adverse events were comparable between the two groups. Conclusions: Adjuvant chemotherapy intensification using HAI did not significantly reduce recurrence following CRLM resection. However, patients with multiple liver metastases, RAS/BRAF mutation and those with positive postoperative ctDNA might benefit from adjuvant HAI treatment. Clinical trial information: NCT03500874.