Based on multiple regression models of reproductive behavior, direct assessment of the impact of the Taiwan family planning program on birth rates is shown to imply different conclusions for policy than those implied by earlier studies focusing on the rates of the adoption of contraceptives. Although its direct impact on birth rates was initially greater than earlier studies have suggested, the program is subject to sharply diminishing returns. Initially, the program did contribute to reduced birth rates among both younger and older women; now, after three years, its contribution is limited to women 30 years and older. Among women in their 20s, birth rates may now be even higher in regions intensively served by the island-wide program. A comparison of the marginal effectiveness of two classes of field workers in Taiwan's family planning program shows that the more productive worker has been underutilized and then dropped from the program. More consistently reliable methods should be developed for cost-effectiveness evaluation of resources allocated to population programs. 71 pp. Ref.