Despite decades of research on racial discrimination in hiring, the drivers of the process are still poorly understood. One simple but still bearing question is whether discriminating agents react to applicants’ phenotypic features in and of themselves, or else take these features as signals of something else (typically applicants’ cultural and/or socio-economic background). Causally identifying the role of phenotype as an independent trigger of discrimination is a challenging task for two main reasons: First, in many countries, phenotype cannot be signalled at all because job applications do not include photographs. This is the case of the US, where most of the discrimination literature comes from. Second, and most importantly, individuals’ phenotype is typically confounded with their ethno-cultural background and, because of accumulated social (dis)advantage, also with their socio-economic background. While socio-economic background can be made orthogonal to applicants’ phenotype in experimental research, making phenotype orthogonal to ethnic background can hamper the external validity of the experiment. To circumvent this problem, we conducted an innovative field-experiment that uses (fictitious) internationally-adopted children in Spain to obtain fully unconfounded phenotypic discrimination estimates. The experiment exploits three unique characteristics of the Spanish case: 1) résumés in Spain generally include a photograph of the applicant, which allows us to signal phenotype; 2) Spain is one of the world-leading countries in international adoptions and this makes adopted children -now entering the labour market- sufficiently common; and 3) Spain has a distinctive two-surnaming system, whereby children get both the father’s and the mother’s surnames, which allows us to signal native ethnicity for both parents. This unique combination of factors provides a credible (i.e. externally valid) pool of ‘multiracial’ natives to test for the distinctive effect of phenotype as a potential trigger of hiring discrimination in real hiring processes. We investigate the responses of roughly 2,000 Spanish firms to an equal number of fictitious Spanish job-seekers applying to 20 different occupations varying in skills and customer contact. All fictitious applicants have the same Spanish-sounding name (Álvaro/Alba Martínez García) but randomly differ in phenotype and country of birth as follows: 20% are born in Spain to biological Spanish parents and have a White phenotype; 80% were born abroad and adopted by Spanish foster parents at 6 months of age (subtly signalled in the cover letter). These adopted children have non-White phenotypes, randomly assigned as follows: 50% Black (adopted in Ethiopia) and 50% Asian (adopted in China). We use the same sixteen validated photographs (eight for each sex) used in Polavieja et al.’s racial discrimination study (2022). Because these photographs were carefully matched in dimensions of attractiveness, competence and sympathy, their potential effects on employers’ hiring decisions can be attributed to phenotypic variation alone and hence provide fully unconfounded estimates of color discrimination. Studying whether non-White adopted children arriving to Spain as infants face discrimination when looking for jobs is also substantively relevant, for it concerns the employment chances of many young Spaniards now entering the labour market.