This paper discusses a number of interventions that are aimed at increasing educational opportunities of homeless students in the United States. Only a few of these programs have been locally evaluated, and the evaluation process is particularly challenging for local evaluators to undertake. At a state and national level, commitment to evaluation for accountability and continuous improvement has not been strong, due to politicization of funding for programs. At the local level, programs rarely know how long students will stay in their program, and they must provide a diversity of services to student of all ages with highly individual needs. This paper considers strategies used by one local homeless education program in Nashville (Tennessee) to address these problems in evaluation process and design. A well-designed evaluation plan was made a component of program planning for the new grant cycle. Program participants were asked to define distant objectives (those for a longer period of time) and nearby objectives (those that could be met with one or two sessions with a student), and then to define strategies for reaching these objectives. Strategies were developed to face the particular obstacles this program faced, obstacles that face most local evaluations of programs. These were: (1) a way to determine the length of stay of subjects and the size of the sample; (2) attempts to improve the accuracy of school records; (3) enhancing the diversity and availability, in terms of location, of services; and (4) better measurement of social development outcomes. State support for evaluation efforts and increased funding and direction at both state and national levels would help the tasks of local evaluators. (Contains 3 tables and 65 references.) (SLD)