Initiated by the seminal work of Diamond/Dybvig (1983) and Diamond (1984), advances in the theory of financial intermediation have sharpened our understanding of the theoretical foundations of banks as special financial institutions. What makes them "unique" is the combination of accepting deposits and issuing loans. However, in recent years the notion of "disintermediation" has gained tremendous popularity, especially among American observers. These observers argue that deregulation, globalisation and advances in information technology have been eroding the role of banks as intermediaries and thus their alleged uniqueness. It is even assumed that ever more efficiently organised capital markets and specialised financial institutions that take advantage of these markets, such as mutual funds or finance companies, will lead to the demise of banks. Using a novel measurement concept based on intermediation and securitisation ratios, the present article provides evidence which shows that banking disintermediation is indeed a reality for the US financial system. This seems to indicate that American banks are not all that "unique"; they can be replaced to a considerable extent. Moreover, many observers seem to believe that what has happened in the US reflects a universal trend. However, empirical results reported in this paper indicate that such a trend has not manifested itself in other financial systems, and in particular, not in Germany or Japan. Evidence on the enormous structural differences between financial systems and the lack of unequivocal signs of convergence render any inferences from the American experience to other financial systems very problematic.