Dual-axis rotational inertial navigation systems (RINSs) can be categorized into inner-azimuth RINSs and outer-azimuth RINSs based on different frame structures. Previous systematic calibration methods mainly involved tri-axis turntables or inner-azimuth dual-axis turntables. However, the conventional calibration methods cannot be applied to outer-azimuth dual-axis RINSs due to the constraints on mechanical structure. Upon analyzing the inertial measurement unit (IMU) error stimulation of outer-azimuth dual-axis RINS under static conditions, specific IMU errors are revealed to be unobservable or weakly observable, which provides insights into the unavailability of self-calibration. To address this issue, an in-flight calibration method based on incremental azimuth observation is proposed. Specifically, we first provide a theoretical analysis in terms of error stimulation to lay a solid foundation for calibration with outer-azimuth dual-axis RINSs. Then, a novel approach based on incremental structure and azimuth information is proposed to tackle the problem of weak observability associated with gyro errors. Moreover, a principle for designing the calibration rotation scheme of IMU is developed and the corresponding 24-position rotation path is further created. Simulations and flight tests demonstrate that the estimations of all error parameters can achieve convergence with the proposed 40-D Kalman filter. By contrasting the maximum positioning errors of 1-h pure inertial navigation tests, it is indicated that the proposed 40-D Kalman filter is found to be equally effective as the conventional 39-D Kalman filter, which fully illustrates the validity and feasibility of the proposed calibration method in outer-azimuth dual-axis RINSs.