This paper discusses query optimization strategies in distributed relational database management systems. It analyzes semijoin and join strategies (rather than specific algorithms) and how they are effected by processing and communication costs. The relative advantage of a strategy is dependent on the various parameters of the problem, which are identified in this paper. The behavior of the strategies in the case of unlimited buffer capacity was tested in simulation experiments. In order to concentrate on the effectiveness of a strategy rather than a particular algorithm, we solved all problems to optimality by enumerating the solution space. It is shown that no one strategy always dominates the others, and consequently, it is important to develop sophisticated optimizers that will adapt to changes in the environment.