This study (the first part of a two-part study) analyzed the effectiveness of three mandatory welfare employment programs in serving different segments of the Aid to Families with Dependent Children (AFDC) caseload. Data were collected in evaluations of welfare employment initiatives in San Diego, Baltimore, and several counties in Virginia. Program participation was required for different portions of the AFDC caseload. Eligible applicants and recipients (primarily female) were randomly assigned to experimental groups, which received program services, or to control groups, which did not. Data were collected using AFDC payments and Unemployment Insurance earnings records. With few exceptions, employment and earnings impacts were consistently smaller than average for the welfare applicants and recipients who had the best work records and the least prior welfare experience. The impacts were usually larger for more dependent individuals, although not for the cases that were the most dependent. Programs had less consistent impacts on subgroups defined by characteristics such as marital status and educational level. The outcome measures examined were not valid indicators of program performance. Neither job entries nor cases off welfare were a satisfactory predictor of the changes in employment, earnings, and welfare receipt achieved by the programs studied. (20 references) (YLB)