Teacher's aides are used worldwide, in various school systems, although, there is no strong evidence of their impact on student outcomes. We use a randomized trial to challenge this state of evidence. We randomly allocate 105 schools to two types of treatments - aides with or without a teaching degree - compared to a control group. Both types of aides have positive impacts on test scores and the effects are persistent over time for disadvantaged students. Exploratory analyses of mechanisms suggest that a teacher's aide is not just a class-size reduction, but especially impactful when sharing instructional responsibility for the classroom.