Micro aerial vehicles may build a collective of smart sensor objects cooperating in civilian missions such as search and rescue, farmland monitoring, or surveillance. Wireless connectivity is a prerequisite for transferring images and other sensor data to the ground. Though aerial vehicles may set up a multi-hop wireless network on their own, vehicle movement causes frequent changes of the wireless signal quality and intermittent connectivity, which poses challenges to end-to-end data delivery and renders traditional routing approaches impractical. We address this problem by including delay-tolerant packet forwarding. Further, we make forwarding mission-aware, i.e., aware of future positions and connection opportunities derived from the waypoints of the MAVs' mission. The resulting path options for packet forwarding open a vast search space. We present a solution to find a path efficiently based on the $A^*$ search algorithm. We study the performance of our mission-aware algorithm compared to a delay-tolerant variant of geographic routing in simulation and in a testbed of quadcopters with IEEE 802.11n aerial links. Our first results reveal that for simple scenarios, the benefit of mission-aware forwarding is limited, yet, in more sophisticated scenarios, mission-aware forwarding can alleviate inefficient forwarding and improve performance.