Proof of the value of bilingual education programs is usually based on evaluations which measure success solely on the basis of student test scores. An alternative method of evaluating such programs as it is implemented in three bilingual education programs in Connecticut elementary schools is reported in this paper. Eleven first- and second-grade Spanish/English classrooms containing a total of 257 students were visited twice a month for four months in 1986 to identify and describe the administrative, curricular, and instructional practices that contribute to program success. These characteristics were identified through qualitative methods such as classroom observations, interviews, and reviews of curricular materials. Qualitative data were then linked to students' test scores to provide a more accurate measure of program success than test scores alone could have provided. Program characteristics identified as contributing to success included: numerous opportunities for students to write; presence of a strong core curriculum in the bilingual education program matching the district's all-English curriculum; a well-defined plan for transitioning students to English reading and writing; highly-trained teachers; access to curricular materials; strong support for the native language as a bridge to learning English; principal support; and integration and acceptance into the mainstream school structure. An 11-item bibliography is included. (MSE)