This paper makes the case that the current single-axis approach to the diagnosis and remedy of pay discrimination is inadequate in the case of multiple disadvantage. While a good deal is known about pay gaps, particularly those affecting women, less is known about those affecting people in other disadvantaged groups and those in more than one such group. This analysis of multiple years of pay data, n = 513,000, from a large UK-based company shows that people with more than one disadvantaged identity suffer a significantly greater pay penalty than those with a single disadvantage. The data also suggest that penalties associated with multiple disadvantage exponentially increase. In other words, disadvantages seem to interact to the detriment of people at ‘intersections'. The paper considers the implications for policies aimed at reducing pay inequalities. These currently take a single-axis approach and may be misdirected.