Despite Brazil’s progress in labor market outcomes over the last decade, labor market outcomes of youth – particularly those from disadvantaged socio-economic contexts – continue to be markedly worse than almost any other demographic group. Youth face lower wages, higher levels of informality, and more frequent unemployment spells. Their attachment to the labor market is also tenuous and irregular, with frequent entry and exit. This paper analyzes an innovative approach at dealing with unemployment among disadvantaged youth living in Rio de Janeiro’s slums (or favelas), pioneered by a small NGO, Galpão Aplauso. The novel dimension of this program is the use of expressive arts and theatre as a pedagogical tool. This tool is used as part of a training program that also includes vocational and academic training, as well as training in work-readiness skills. We expect that the bundle offered by Galpão may impact employment through a number of channels. First, it can teach specific academic and vocational skills which employers demand. These are basic numeracy, reading, as well as specific skills required by certain trades. Second, it can teach specific conducts required to work in a formal job environment. These work-readiness skills are conducts that employers expect from the workforce, but that youth in the marginalized communities in Rio’s favelas often lack. Beyond the specific technical or work-readiness skills, the program can build broader cognitive and socio-emotional skills (sometimes referred to as non-cognitive skills) that are known to be important for labor market outcomes. The economics and psychology literature shows that socio-emotional skills do change over the life-cycle and can me molded, particularly at younger ages. However, whereas specific work-readiness skills may be easier to change, we are more ambivalent regarding the ability of youth-training programs to impact broader types of socio-emotional skills for youth and young adults. Program participants could see higher labor market outcomes due to formation of vocational, academic, and socio-emotional skills. Using standard labor market participation questions, we can test the program’s overall impact on employment and earnings. We can also test the program’s impact on broad socio-emotional skills measured by standard psychological tests. However, we cannot parse out what share of the program impacts is due to higher vocational skills, higher academic skills, or higher socio-emotional skills, as the experiment was not designed with separate treatment arms that would allow for separate estimation of these different channels.