Aim: Clinical features of esophageal cancer (EC) patients have poor prognostic power. Thus, it is paramount to discover biomarkers that can allow a more accurate survival prediction. Methods: To detect genetic variants associated with survival, DNA from 120 patients treated with cisplatin-based neoadjuvant therapy were genotyped using drug metabolism enzymes and transporters array. Results: We identified two variants: the rs2038067 in PPARD (p = 0.0004) and the rs683369 (F160L) in SLC22A1 (p = 0.001). Their prognostic power was greater than that of clinical stage alone (p = 0.017) and comparable to that of response to neoadjuvant therapy (p = 0.71). Interestingly, the prognostic accuracy of response models increased significantly when genetic variables were included (p = 0.003). Conclusion: Our data, though preliminary, strengthen the potential utility of germline variants for a better-tailored management of EC patients.