Based on data from the 1982 National Survey of Family Growth, exposure to the risk of unintended pregnancy is classified by use of specific contraceptive methods and by nonuse, and average rates of unintended pregnancy are estimated for each type of exposure. Three hypothetical models of improved contraceptive practice are then applied to the data for all women and for age, race and marital-status subgroups. The first two models assume increases in the use of some existing contraceptive methods, but only the second model additionally assumes the introduction of new methods. The third model assumes the complete elimination of nonuse of contraception. These models yield different estimates of the reduction in unintended pregnancy rates which are illustrated for various age-groups, for whites and blacks and for married and unmarried women. Among all women aged 15-44, the changes assumed by Model I imply a 32 percent reduction in unintended pregnancy; Model II implies a 56 percent reduction; and Model III implies a 57 percent reduction. The implied reductions in abortion are in a similar range.