In this paper we describe a unique three-dimensional bin packing problem with non-convex objects having cavities and holes. Part packing takes place in an environment where parts float as they would in a weightless environment. The application domain is the selective laser sintering rapid prototyping technique. A genetic algorithm is used as a search method to find a good part-packing solution. The fitness evaluation for the genetic algorithm is based on the actual part geometry as described in the Stereolithography (STL) file. Part-intersection detection utilizes several methods common in computational geometry. Initial results are promising, showing that the genetic algorithm is able to find a good solution for such a difficult packing problem.