Merit pay systems based on output may fail because establishing valid standards for productivity that account for educational goals, individual student needs, and the value of cooperative efforts is extremely difficult. Systems based on evaluations may fail if employees are not convinced that the systems lack bias. Research on the characteristics of successful merit pay systems was conducted in 6 public school districts that had used merit pay for at least 5 years and either used pay differentials of at least $1,000 or served over 10,000 students. Only 7 of 11,500 school districts originally surveyed in 1978 met these criteria in 1983, out of 47 that reported using merit pay. No urban districts met the criteria. The strategies used by the six districts to make merit pay acceptable included providing extra pay for extra work, making everyone feel special, keeping the program inconspicuous, and legitimizing the program through employee participation. The research findings suggest that the uses of merit pay in education may not differ significantly from those in other industries, that workers' responses to merit pay systems may be sensitive to factors other than the merit incentives themselves, and that workers' participation in system development and revision is critical. (PGD)