This paper provides some empirical evidence on the link between computer use and the efficiency and effectiveness of voluntary, direct users of personal computers. A survey of accounting professionals in the Internal Revenue Service (N = 1110) provides self-reported levels of use, efficiency and effectiveness, as well as data on training, management policy and user characteristics. A second survey of completed audits (N = 1851) provides an objective standard of comparison for the self-reported use and performance data. Although users believe the computer is faster for some tasks, it does not improve their overall efficiency. The divergence is accounted for in part by managerial policies that encourage use of the system for marginal tasks and in part by users choosing to use the tool for activities that could be done more quickly manually. Data on the association between use and effectiveness show that while users believe the computer makes them more effective, much of this perceived value is symbolic rather than substantive. Users benefit from a sense of professionalism and self-esteem, but it is not clear whether the organization as a whole benefits. The low association between use and productivity suggests that researchers should resist the temptation to regard use as a proxy for implementation success in the absence of actual productivity measures. Practitioners should be aware that policies which promote use may actually hurt productivity by encouraging users to apply technology to tasks where it is only marginally useful.