A quasi-experimental study compared achievement effects of various configurations of the Success for All (SFA), schoolwide reform model for elementary schools serving many at-risk students, from full implementation (certified tutoring, full-time school-level facilitator, and the family support team) to implementations lacking some elements felt to be essential. Full implementation of both the English and Spanish configurations of SFA in the Houston Independent School District (Texas) was achieved during the second year of the program, 1995-96. Two student cohorts participated in the English program--595 students who began first grade in 1995 and 682 students who began first grade in 1996. Subjects were in 46 SFA schools and 18 comparison schools. The final sample of subjects in the Spanish program consisted of 278 first-grade students in 20 SFA and 10 comparison schools. The Language Assessment Scales and a battery of 4 reading posttests were administered. Results indicated that: (1) programs with part-time facilitators, certified tutoring, and a family support team produced positive effects in moderately impoverished schools, but not in extremely impoverished schools; (2) schools that did not make the basic organizational changes required to provide family support and certified tutoring performed worse than comparison schools; (3) a full-time facilitator was essential in extremely high-poverty schools serving predominantly African American student populations; and (4) schools that provided certified tutoring achieved better results in the English language program. Findings suggest that schools should undertake implementation of the whole program, and that comprehensive changes in teachers' daily practices and in school organization are needed to bring about major changes. Contains 16 references, 6 tables and 3 figures of data.) (RS)