Background <p>Treatment options for esophageal cancer (EC) are increasingly diverse and complicated. We conducted a web-based survey on patient preferences regarding systemic drug treatment for EC in Japan.</p> Methods <p>We used a discrete choice experiment to determine patients’ preferences. Eight relevant attributes and their levels were determined using step-by-step input from patients and EC medical experts. Four attributes were related to efficacy, three to safety, and one to quality of life. Ten choice sets of two hypothetical treatments with independent attribute levels were presented in questionnaires. A multinomial logit model was used to estimate predicted choice probabilities. We calculated means and 95% confidence intervals of preference weights and relative attribute importance (RAI). The primary endpoint was mean RAI; secondary endpoints were attribute trade-offs for the total sample, and mean RAI and attribute trade-offs for subgroups. Eligible patients with EC (undergoing or having undergone treatment) were recruited through commercial panels.</p> Results <p>We analyzed 149 response sets. Respondents placed the highest relative importance on 1-year overall survival (OS; 31.4%), followed by hospitalization/dosing time (27.3%). Safety attributes, including immune-related adverse events, had relatively little influence (≤ 7.5%). Patients were willing to trade off 17.4% of 1-year OS by changing from hospitalizations and long dosing time to no hospitalization and short dosing times. The subgroup aged ≥ 65&#xa0;years placed greater importance on quality of life than survival.</p> Conclusions <p>We first clarified patients’ preferences for EC systemic therapy in Japan, which could provide useful information in EC treatment selection.</p>