End-user statistical computing is one way to improve the productivity of researchers; it does so by providing them with a statistical computing environment and with guidance on its appropriate use. Critical success factors include: (1) ease of use; (2) software correctness and robustness; (3) compatibility with system resources; (4) training and education; (5) short-term and long-term consultation; and (6) management sponsorship. Researchers need to know how to use a statistical method; they do not need detailed knowledge of a statistical method's underlying mathematics. With appropriate guidance and contact, researchers can directly control the execution of a statistical method in an appropriate manner. The social payoff is that more scientists, engineers, and technicians can design and analyze experiments with shorter turnaround times than without end-user statistical computing. Future directions involve expert system components of the user interface and incorporation of computer-intensive methods. A 59-item list of references is provided. (SLD)