Abstract Head and neck squamous cell carcinoma (HNSCC) patients are at an increased risk of developing a second primary tumor (SPT) or recurrence following curative treatment. 13-cis-retinoic acid (13-cRA) has been tested in chemoprevention clinical trials, but the results have been inconclusive. We genotyped 9,465 single nucleotide polymorphisms (SNP) in 450 patients from the Retinoid Head and Neck Second Primary Trial. SNPs were analyzed for associations with SPT/recurrence in patients receiving placebo to identify prognosis markers and further analyzed for effects of 13-cRA in patients with these prognostic loci. Thirteen loci identified a majority subgroup of patients at a high risk of SPT/recurrence and in whom 13-cRA was protective. Patients carrying the common genotype of rs3118570 in the retinoid X receptor (RXRA) were at a 3.33-fold increased risk (95% CI, 1.67–6.67) and represented more than 70% of the study population. This locus also identified individuals who received benefit from chemoprevention with a 38% reduced risk (95% CI, 0.43–0.90). Analyses of cumulative effect and potential gene–gene interactions also implicated CDC25C:rs6596428 and JAK2:rs1887427 as 2 other genetic loci with major roles in prognosis and 13-cRA response. Patients with all 3 common genotypes had a 76% reduction in SPT/recurrence (95% CI, 0.093–0.64) following 13-cRA chemoprevention. Carriers of these common genotypes constituted a substantial percentage of the study population, indicating that a pharmacogenetic approach could help select patients for 13-cRA chemoprevention. The lack of any alternatives for reducing risk in these patients highlights the need for future clinical trials to prospectively validate our findings. Cancer Prev Res; 4(2); 185–93. ©2011 AACR.