Two studies were designed to extend Butler's (2007) model and measure of achievement goals for teaching, to recognize that teaching is an interpersonal endeavor, not just personal endeavor. In Study 1, results from 530 teachers in Israel confirmed the predicted 5-factor model comprising relational goals, whereby teachers aspire to create close and caring relationships with students, in addition to the previously identified mastery, ability-approach, ability-avoidance, and work avoidance goals for teaching. Results from a subsample of 272 teachers confirmed that the teachers' goals were coherently and differentially related to their teaching practices, assessed several months later. The most important results showed that only relational goals predicted teacher social support; they also predicted mastery instruction. Teacher ability-approach and ability-avoidance goals both predicted performance instructional practices. Multilevel analyses of data from 73 teacher-class pairs (1,790 students) in Study 2 showed that teacher relational goals also predicted student reports of teacher social support and mastery instruction; mastery goals were negatively associated with student perceptions of performance practices. Teachers' goals, but not teachers' reports of instruction, predicted students' perceptions of instruction. The results supported extension of Butler's model to incorporate relational goals for teaching and confirmed that strivings to connect are at the heart of effective teaching. The results also shed new light on relations between teachers' goals and teachers' approaches to instruction and on teacher and student perspectives on instruction. Differences in boys' and girls' perceptions of instruction imply that it is important to consider possible gender differences in research on classroom goal structures. (Contains 8 tables and 3 figures.)