Background: We aimed to evaluate the clinical efficacy of adding temozolomide (TMZ) to preoperative capecitabine (CAP)-based chemoradiotherapy (CRT) in patients with locally advanced rectal cancer (LARC) and validate O⁶-methylguanine DNA methyltransferase (MGMT) methylation status as a predictive marker for TMZ combined regimens. Methods: : LARC patients with clinical stage II (cT3-4N0) or III (cT_anyN+) disease were enrolled. They were stratified into MGMT unmethylated (uMGMT) and MGMT methylated (mMGMT) groups by methylation-specific PCR before randomization, and then were randomly assigned (1:1) to one of four treatment arms: uMGMT/CAP (arm A), uMGMT/TMZ+CAP (arm B), mMGMT/CAP (arm C), and mMGMT/TMZ+CAP (arm D). The primary endpoint was the pathologic complete response (pCR) rate. Results: Between November 2017 and July 2020, 64 patients were randomized. Slow accrual caused early study termination. After excluding 4 ineligible patients, 60 were included in the full analysis set. The pCR rate was 15.0% (9/60), 0%, 14.3%, 18.8%, and 26.7% for arms A, B, C, and D, respectively (p= 0.0498 between arms A and D). The pCR rate was 9.7% in the CAP group (arms A+C), 20.7% in the TMZ+CAP group (arms B+D), 6.9% in the uMGMT group (arms A+B), and 22.6% in the mMGMT group (arms C+D). Grade 1–2 nausea or vomiting was significantly more frequent in the TMZ+CAP treatment groups (arms B+D) than in the CAP treatment groups (arms A+C, p< 0.001) with no difference in grade 3 adverse events (AEs). There were no grade 4 or 5 AEs. Conclusions: The addition of TMZ to CAP-based CRT tended to improve pCR rates, particularly in those with mMGMT LARC. MGMT status may warrant further investigation as a predictive biomarker for chemotherapeutic agents and radiotherapy. Clinical trial information: NCT03156036.