We investigate the role of physical appearance, in addition to race and sex, in the rate of discrimination observed in the labour market of Lima. Our experimental design allows us to disentangle the effect of each of those three variables on the callback rates received by our fictitious job candidates. Since we are controlling for variables that are important in the selection process (mainly, education and job experience), our results provide better indicators of discrimination than the ones we could obtain through the econometric analysis of observational data. We find that discrimination based on looks is greater than that based on race or sex. The first two types of discrimination are in professional and unskilled jobs.