Typically, recognition of other-race faces is poorer than recognition of own-race faces. This ‘other-race effect’ can have devastating consequences. For instance, a high proportion of wrongfully imprisoned people who were later exonerated by DNA evidence were convicted on the basis of misidentification by other-race eye-witnesses. In contrast to these real-world cases of complete failure to recognise other-race people, the other-race effect found in the lab is generally relatively small. To investigate this apparent paradox, we took an individual differences approach and found considerable variability in the size of individuals’ other-race effects (Wan, Crookes, Dawel, Pidcock, Hall, & McKone, 2017). Importantly, we identified a new group (approximately 8% of the population) who were so poor at other-race face recognition, that they met the criteria for clinical-level impairment (i.e., they were ‘face-blind’ for other-race faces). The risk factors for other-race face blindness included having low contact with people from the other-race and being at the lower end of the normal range for own-race face recognition. However other-race face-blindness was not associated with applying less effort for other-race than own-race faces. This discovery answers a central theoretical question as to whether poorer performance with other-race faces is primarily due to expertise or to motivational biases (e.g., prejudice), and highlights a critical role for experience. The fact that nearly 1 in 10 people are very poor at recognising other-race faces also has major implications for everyday life, not only for the important case of eyewitness testimony, but also for everyday social interactions, where some people will struggle to recognise acquaintances and colleagues from an unfamiliar race.