Aid ineffectiveness, fragmentation, and volatility have already been highlighted by scholars and OECD studies. Far fewer studies have been devoted to another problem of capital flows: herding behaviour. Building upon a methodology applied to financial markets, where herding is a common feature, this article attempts to measure herding behaviour in the allocation of foreign aid, proposing different indexes that try to capture the specific features of aid allocation. Of course, herding can also be beneficial. When a country faces an earthquake, a tsunami, or any humanitarian disaster, the rush of donors is a positive factor. Excluding such cases of beneficial herding, we attempt to focus on pure herding behaviour, creating pendulum swing effects comparable to those in financial markets. . Our different indexes all detect donor herding, its exact size depending on the measure adopted. Our preferred index, relying on threeyear disbursements, indicates a significant level of herding, similar to that which is found on financial markets. We also uncover major differences across different types of donors, with no, or very limited, herding among multilateral donors, in contrast to bilateral donors, always subject to herding behaviour. We then follow by investigating the empirical causes of herding. We find that while political transitions away from democracy are accompanied by herding out, transitions towards democracy do not affect herding levels. Finally, we show that observable determinants actually explain little of the herding levels, leaving a large part of herding unexplained.