This article investigates the biases involved in estimating the effects of teacher attributes on school’s performance. The study was performed for Chilean educational system, where student distribution is differentiated by schools and teachers are not randomly assigned to them. Findings showed that teacher attributes which favored learning appeared more frequently alongside higher socioeconomic status students. When correcting the bias, results showed that the effects of teacher attributes have been overestimated for the vast majority of characteristics. Nonetheless, attributes such as teaching experience, being a woman, having short-term specific professional training, and having a greater curriculum coverage continued to have positive impacts on the performance of 4th grade students.Este artículo investiga el sesgo presente en las estimaciones del efecto de los atributos docentes en el desempeño escolar en el sistema educativo chileno, donde estudiantes y profesores no son aleatoriamente asignados a las escuelas. Los resultados señalan que los estratos socioeconómicos altos presentan con mayor frecuencia atributos docentes que promueven el aprendizaje. Cuando se corrige por este sesgo, los resultados mostraron una sobrestimación del impacto de la gran mayoría de atributos docentes. Sin embargo, atributos como la experiencia docente, ser mujer, con capacitación profesional específica y tener una mayor cobertura curricular, continúan siendo atributos que impactan positivamente en el desempeño escolar.