The great improvement of algorithms and computing hardware in the last few years must be ranked as one of the most important turning points in the history of multiphase flow research. After a brief review of some of this recent progress, it is pointed out that, besides its application to solving actual problems, computational physics plays other key roles: (1) As a tool to develop and understand basic physics and as a guide toward asking more penetrating questions; (2) As an aid in closing the averaged equations; (3) As a means to learn to compute better. Roadblocks toward greater effectiveness are the huge complexity of many of the necessary computational tasks but also, at a more practical level, the transmission of "computational knowledge" from one researcher to another, much in the same way as experimentalists can rely on readily available equipment (e.g... lasers, etc.), without having to build each item themselves. The solution to this problem will require a cultural shift-from a "cottage industry" to a "big science" mentality-which can be aided by a different attitude on the part of the funding agencies. Great synergism can be achieved by a closer integration of the multiphase computational physics enterprise with both Applied Mathematics and Computer Science. (C) 2003 Elsevier Ltd. All rights reserved.