ments and research centers than in community-based schools and organizations. The prevention strategies evolved from an emphasis on individual change to an emphasis on policy and environmental interventions. These demonstrations illustrate both the value of building system capacity (there was a substantial need for educational curricula) and the challenges of collaborating with community groups to test applications. The replications enhanced curriculum development but appear to have contributed little to science.

The 1987 Stewart B. Mckinney Homeless Assistance Act (P. L. 100-77) authorized initiatives to address the national problem of widespread homelessness. NIAAA and NIDA collaborated to support demonstration projects that implemented and evaluated interventions for homeless men and women with alcohol- and drug-related problems. Initially, nine projects were funded in eight cities. Each project was required to allocate at least 25 percent of the award for process and outcome evaluation. A separate contract was awarded for cross-site evaluation, coordination, assistance, and data analysis (Lubran, 1990; Orwin et al., 1993). A diversity of interventions was encouraged because there was little empirical data on effective services for alcohol and drug involved homeless individuals (Huebner and Crosse, 1991). The applicants tended to be community organizations or state or local health departments. The community organizations subcontracted with academic-based investigators for the evaluation research. Each site was unique. A special issue of the Alcoholism Treatment Quarterly (McCarty, 1990) and reports from NIAAA (Murray, 1993; Shane et al., 1993) provide more details.

The evaluation report on the first round of demonstrations drew lessons, noted key findings, and made recommendations (Orwin et al., 1993). An obvious but often overlooked finding was that when working with homeless men and women, issues related to food, shelter, and security must be addressed before treatment can be initiated. Programs also learned that both program structure and flexible responses were necessary to engage and retain homeless participants. Start-up required substantial resources and persistence, especially when there was resistance to siting services in specific locations. Overall improvements in client functioning were modest. Generally, the services led to reductions in alcohol and drug use. Composite scores from the Addiction Severity Index suggested improvements in employment and economic security in some of the study sites. Housing stability was increased in a project that facilitated access to alcohol- and drug-free housing; psychiatric status improved in a different city.

Substantial project variation made cross-site comparisons difficult, and