Current and future space observation missions need to perform many large-angle, multi-axis slew maneuvers between observations while keeping the scientific instrument’s attitude in a safe region. The state-of-practice typically divides each multi-axis maneuver into a series of single-axis sub-maneuvers, each of which is computed by restricting its guidance solution to the exact spacecraft momentum capacity. This ensures that the constraints are explicitly considered and results in a simple on-board implementation of the guidance algorithm, but is time-consuming and non-optimal for the whole multi-axis maneuver. Addressing this issue, this article presents a novel analytical guidance approach that relies on the convexity of the permissible attitude zone. The proposed guidance is time-optimal for a given spacecraft design and set of admissible observation targets. Both guidance approaches are compared using a multi-body/multi-actuator benchmark spacecraft, whose complex repointing phase requires an autonomous on-board guidance computation. It is shown that the proposed approach is systematic and that the reduction in maneuver time, compared to the state-of-practice approach, is considerable.