This report seeks to provide a portrait of schools serving rural communities in northwest China, and to shed light on factors that encourage and discourage school persistence among children in this region. To achieve these goals, we analyze a survey of rural children and their families, schools, and teachers in Gansu province. The project interviewed children in the year 2000, when children were 9 to 12 years old, and again four years later. In part one of the paper, we provide a descriptive overview of the material, human, and cultural resources available in sampled primary and middle schools. Where possible, we note changes between 2000 and 2004. We describe the following types of resources: (1) basic facilities; (2) financial arrangements; (3) teachers, including their background, qualifications, working lives, professional development activities, satisfaction with work, and attitudes about school management and culture; and (4) classroom environments, as reported by teachers and by students. In this descriptive section of the paper, we highlight basic infrastructure issues, the complexity of financial arrangements at the time of the surveys, problems of teacher wage arrears and teacher morale, and the pedagogies and learning environments in classrooms, as reported by teachers and students. In part two of the paper, we investigate reasons for school leaving reported by village leaders, families, and children themselves, and analyze contributors to subsequent enrollment, change in attainment, and attainment of nine years of compulsory education. Our models of family, teacher, and school effects on outcomes show that higher socio-economic status children are more likely to show grade attainment, continued enrollment, and attainment of nine years of basic education. In contrast, the gender story is mixed: girls are less likely to be enrolled, but have not gained less grades, nor are they less likely to achieve nine years of education. This finding suggests that boys may start later or repeat more. It is possible that boys are more likely to be encouraged to repeat a grade to complete it successfully or to increase high school exam scores. One significant finding is that the introduction of school and teacher effects, by and large, does not explain away the advantages of children in better off families. School and teacher effects do not consistently matter across the three outcomes. Some interesting findings include that teacher absenteeism in 2000 is associated with less attainment between 2000 and 2004; children with better-paid home room teachers are more likely to attain nine years of school; and children in schools with minban teachers are less likely to attain nine years. However, there is not a consistent story of school characteristics that help or hinder childrenʹs persistence. Reports by village-leaders, fathers, mothers, and children themselves indicate that, along with socioeconomic status, children's performance and engagement are significant factors in school continuation decisions in Gansu's rural villages. Multivariate analyses indicate that childrenʹs early aspirations and performance matter for later outcomes. We close by discussing the most significant strengths and weaknesses identified among the school resources discussed in part one, and the most significant supports and hindrances to favorable educational outcomes considered in part two.