This paper addresses the problem of finding the minimum time rendezvous point for a geographically distributed group of heterogeneous mobile robots. In contrast to the traditional treatment of the multi-agent rendezvous problem, focus is given mainly to the identification of the globally optimal solution rather than the behavior of the system based on a given control policy. Level sets are introduced as a tool to solve this problem by first computing an arrival time map for each robot, subject to speed, terrain, and dynamic constraints. The computation is parallelizable by requiring each agent to generate its own arrival time map. The arrival time maps can be easily combined to give the overall minimum time rendezvous point. Despite the apparent simplicity of this approach, it is capable of accommodating numerous complicating factors with minimal modification while simultaneously generating a target path trajectory for each robot through the state-space. Examples involving ground, sea, and air robots are used to illustrate the power of this technique.