Ethnic minority underachievement remains a problem in many countries, and affirmation interventions offer a promising approach to help reduce the achievement gap. We compared the effects of a conventional self-affirmation intervention with a dual-identity affirmation on test performance in a minority-concentrated school (56% Black pupils), in London, UK (N=179, M_age=12.29). A randomized design consisted of a new dual-identity condition, a traditional self-affirmation condition, and two control conditions—a ‘one-group’ condition and a non-affirmation control condition. Teachers implemented the interventions in class, and test performance was the outcome measure. As expected, we found Black pupils outperformed non-Black pupils when they undertook a dual-identity affirmation exercise, while non-Black pupils outperformed Black pupils in the traditional self-affirmation condition. Stereotype threat partially mediated this effect: Dual-identity was less threatening for Black pupils than for non-Black pupils, increasing the test performance of Black pupils. We propose dual-identity affirmation as a promising affirmation intervention to reduce threat and improve performance for underachieving minorities in ethnically diverse settings. Implications for teachers as key players in affirmation interventions are discussed.