Using administrative data from linked private schools from one of districts in India that matches 8,319 pupils to their subject specific teachers at the senior secondary level, we estimate the importance of individual teachers on student outcomes in the high-stake senior secondary exam (at the end of twelfth-grade). We control for prior achievement at the secondary level (at the end of tenth-grade) to estimate the value added of a teacher over a two year course, and define a good teacher as one who consistently gets higher achievement for students. In addition to the prior achievement, we exploit the fact that students took exams in multiple subjects during their senior secondary exam to control for pupil fixed effects. We find a considerable variability in teacher effectiveness over a two year course – a one standard deviation improvement in teacher quality adds 0.38 standard deviation points in students score. Furthermore, consistent with studies in the US, we find that although teacher quality matters, the observed characteristics explain little of the variability in teacher quality.