Background: Multiple studies found virtual visits (real-time video or telephone calls with a medical provider) were not equitably accessible to all people during the COVID-19 pandemic. We explored structural barriers to virtual visit access by evaluating disparities in the socio-demographic characteristics of adult oncology patients who reported being offered a visit format choice during appointment scheduling. Methods: We used electronic health record data to identify adults aged ≥21 treated for cancer within the last three years at an academic medical center or affiliated community practice. Patients with a scheduled oncology-related appointment type that could be eligible for virtual format were recruited via a letter of study introduction followed by telephone call(s) to complete a brief survey prior to and following their scheduled appointment. We approached all patients scheduled for a virtual visit and weekly random samples of those scheduled for in-person visits, oversampling Black adults. Participants were recruited between 04/04/22 and 1/26/23. We used logistic regression to evaluate disparities in who on the pre-visit survey reported being offered a visit format choice. For each patient characteristic evaluated, we present unadjusted and adjusted estimates. Covariates in adjusted models were selected in consideration of causal pathways, including structural mechanisms (e.g., racism, sexism). Results: N = 412 patients completed the pre-visit survey (23% response rate), 16% prior to a scheduled virtual visit. 38% of respondents self-identified as Black, 67% as female, 42% as not married; 14% reported difficulty getting by with their current household income, and 51% as not having a college education. Respondents had a mean age of 63 years (SD = 13), and 31% rated their health as excellent/very good, 40% as good, and 29% as fair/poor. 18% reported being offered a visit format choice during appointment scheduling. In unadjusted analyses, we did not find statistically significant differences in patients who reported being offered a visit format choice by race (16% Black, 20% non-Black), education (19% no college degree, 18% college degree⁺), marital status (18% not married, 19% married), personal finances (19% difficulty, 19% no difficulty), and self-reported health status (15% poor/fair, 21% good, 19% very good/excellent). We found significant unadjusted differences by gender (15% female, 24% male) and age (younger patients more likely to report choice). Only male gender remained significant after covariate adjustment (OR = 1.76, 1.03-3.00). Conclusions: Almost a fifth of patients receiving cancer care were offered a visit format choice. Apart from gender, no socio-demographically defined categories of people reported being offered virtual visits less frequently than others. Other structural contributors to virtual visit accessibility, including reliable internet and device access, warrant study.