Context: The literature on mindsets is now well-established in showing the impact of student interventions on their school achievement (Blackwell et al., 2007). But there is much less evidence on the effects of mindset interventions on teachers. Studies have analyzed changes in teacher beliefs (Deemer 2004; Rattan, Good, and Dweck 2012; Rubie-Davies et al. 2015; Claro 2016); however, little is known about the impact of such changes on pedagogical practices (Sun 2018). Objective: This research project aims to answer two questions: (1) Can we change teachers' beliefs in a socially vulnerable context? (2) How do these changes affect teachers' pedagogical practices? To answer these questions, we did a randomized controlled trial of workshops to teachers aimed at demonstrating how can growth mindset be translated to the classroom. Intervention: The intervention was a structured workshop of 5 weeks to 5th-grade teachers, applied at the beginning of the school year in Rio de Janeiro, Brazil. Rio has a high inequality rate in student performance results and is also one of Brazil's most violent cities. The content of the workshops aimed at building the belief that, just like a muscle, intelligence can be increased through effort, coping with adversity, and permanent efforts to improve. The workshops included discussions about brain functioning and its relation to the growth mindset. The workshops also discussed the concept of "stereotype threat" in vulnerable contexts (Steele and Aronson 1995). We hypothesize that, in promoting changes in intelligence beliefs, the implementation of mindset workshops helps teachers in their pedagogical practices to improve student learning. Research Design: We ran a two-level clustered randomization with treatment at the school level. Enrollment in the research meant that both 5th-grade teachers in a school with two classes accepted to participate. The randomization was done by pairing, using the predicted value of the students' results inferred from the schools' characteristics in previous years. Out of the 178 schools that enrolled to participate, we randomly selected 50% of the sample, or 89 schools, to receive the treatment and 50% to the control, totaling 272 teachers with complete data. Data Collection and Analysis: Endline surveys applied at the end of the 2019 school year measure transformations in teachers' beliefs resulting from the mindset workshops. An innovation of this research is that we also use direct classroom observations to measure changes in teachers' pedagogical practices resulting from transformations in their beliefs. We use a standardized classroom observation instrument, called TEACH+, which combines measurement of teacher time on task, pedagogical practices, and student engagement with measures of instructional quality organized in three aspects: (i) Classroom culture, (ii) Quality of Instruction, and (iii) Socioemotional Skills (Molina et al. 2018). To answer the first research question -- if the intervention impacted the teachers' mindset -- we estimate equation (1) below. We include in the equation controls for each pair of schools, the observer, the teacher's characteristics, the school, and the students of the 5 -grade classes of the elementary school. We also include a control as to whether the teacher gives responses that are "socially desirable." (1) Teachers_Growth_Mindset[subscript kj]=[beta][subscript 0]+[beta][subscript 1] Treatment_school[subscript k]+[beta][subscript 2] Social_Desirability[subscript kj] +[beta][subscript 3]Pair_FE[subscript k]+[beta][subscript 4] Coder_FE[subscript k]+[beta][subscript 5]Teacher_charact[subscript kj]+[beta][subscript 6] School_size[subscript k]+ [beta][subscript 7] Student_charact[subscript kj]+[epsilon][subscript kj]. In the second research question analysis -- if the teachers' mindset impacted their pedagogical practices -- we use an instrumental variable methodology. That is, we use the randomized treatment as an instrumental variable for variations in teachers' mindsets. Thus, we performed a two-stage estimation, as indicated in equation (2) below. (2) First-stage: Teachers_Growth_Mindset[subscript kj]=[beta][subscript 0]+[beta][subscript 1] Treatment_school[subscript k]+[beta][subscript 2]Social_Desirability[subscript kj]+[beta][subscript 3]Pair_FE[subscript k]+[beta] [subscript 4]Coder_FE[subscript k]+[beta][subscript 5]Teacher_charact[subscript kj]+[beta] [subscript 6]School_size[subscript k]+[beta][subscript 7]Student_charact[subscript kj]+[epsilon][subscript kj]. Second-stage: TeachPlus[subscript kj]=[beta][subscript 0]+[beta][subscript 1kj]+[beta][subscript 2]Social_Desirability[subscript kj]+[beta][subscript 3]Pair_FE[subscript k]+[beta][subscript 4]Coder_FE[subscript k]+[beta][subscript 5]Teacher_charact[subscript kj]+[beta][subcript 6]School_size[subscript k]+[beta][subscript 7]Student_charact[subscript kj]+[epsilon][subscript kj]. Results: We observed that the intervention had a positive and significant impact on teachers' growth mindset. Our results indicate that teachers in treatment schools had an increase of 0.311 standard deviations in the growth mindset's standardized indicator. For the second question, we estimate that the TEACH+ average, which refers to a general measure of teaching practice quality, has also significantly increased. Teachers who reported growth mindset beliefs had higher averages in the TEACH+ indicator of teaching quality by 0.528 standard deviations. Regarding the first aspect of the TEACH+ measure (Classroom culture), we observed that teachers who believe intelligence is not fixed have significantly greater averages of classroom culture. This result means, for example, that the teacher explains more often to the student what is expected of the student as "good" behavior. We also find a positive and significant effect of a higher growth mindset in instruction quality: we observe that teachers who believe that intelligence is not fixed more often encourage students to work on critical reasoning. However, this improvement (of 0.478 standard deviations) is statistically significant only at 10%. Finally, we did not find any significant effect of the teacher growth mindset in the indicators of socioemotional skills, such as autonomy, perseverance, and social and collaborative skills. Conclusion: Our results show that we impacted teacher's mindsets and that such changes have significantly affected teachers' pedagogical practices. We observed that teachers' pedagogical practices have changed considering the so-called "classroom culture" and "quality of instruction." These results indicate that teachers with reported growth mindsets treat students with more respect, define more clearly the behavior expectations, and recognizes the students' positive behavior more often. Changing the classroom culture is particularly important in a vulnerable context such as Rio de Janeiro, Brazil, where examples of violent behaviors are familiar to students in the limbo of studying versus going to criminal activity. Additional analyses have also shown the importance of such changes in pedagogical practices to student achievement.