Fischer, Drake, & Breakey, 1992). However, it is also clear that these individuals constitute a group with a multiplicity of problems-of critical intensity and requiring attention. A substantial amount of public funding from federal and state sources has been invested in attempted solutions to the problems of homelessness and mental illness. However, the number of descriptive reports in the literature greatly exceeds that for reported evaluations of these programs. Indeed, although general principles for serving persons who are homeless and mentally ill have been articulated Blankertz, Cnaan, White, Fox, & Messinger, 1990;Sheridan, Gowen, & Halpin, 1993;Susser, Goldfinger, & White, 1990) Levine, 1991;Fischer, 1989;Johnson, 1989;Ribisl et al., 1996) (1989, 1991) Bybee, Mowbray, & Cohen, 1994).
Highlights of Program ResultsPreviously published studies have described the heterogeneity of the population served ). An implementation analysis conducted on the project indicated that in its fully operational period, it was serving the intended population by site and referral source (Mowbray, Cohen, & Bybee, 1991). The project achieved a relatively high success rate in engaging clients screened eligible for services, in that 73 % overall accepted some form of project assistance . This compares favorably to other reports in the literature (Barrow et al., 1989;Rife et al., 1991 . There were no significant differences in rates of engagement by recruitment source at the Collegetown site (see Table 2). For Factorytown, however, the interaction of engagement rate and recruitment source during the implementation period was significant, with marked differences across recruitment sources in the numbers served and in their successful engagement. Thus, we found that during the first screening period from March 1989 through December 1989 (see Table 2), nearly half (46.0%) of the project's eligible clients were recruited from the CMH caseload, and they were engaged with remarkable success (96.6% were fully engaged). However, only a small proportion of the total pool of those eligible were recruited from shelters (7.9%), and only 60% of those recruited were fully engaged. Although there was a good proportion of those eligible recruited from hospital units (46.0%), only 24.1 % of these were fully engaged. These results reflected marked deviations from agreed-on project expectations for 185