Fleet assignment choices influence passenger booking behavior, but conventional robust optimization overlooks this decision-demand dependency, leading to suboptimal revenue performance. This paper addresses this gap by proposing a data-driven contextual robust optimization framework for the fleet assignment problem (FAP) under demand uncertainty. Leveraging historical operational data, we construct a context-dependent demand prediction model in which both fleet assignment variables and ticket prices serve as fixed-design covariates. The demand uncertainty set is parameterized by a budget of uncertainty controlled by a protection level parameter, enabling a tunable trade-off between revenue maximization and robustness against demand fluctuations. The resulting model is formulated as a mixed-integer linear program (MILP). For large-scale instances, we propose a column-and-constraint generation (C&CG) algorithm enhanced with fix-and-optimize heuristics. Numerical experiments on real-world airline data demonstrate that our contextual robust model outperforms deterministic and non-contextual robust benchmarks, improving worst-case revenues by 12–18% with acceptable computational overhead.