Summary: Terminological logics (TLs) are knowledge representation formalisms of considerable applicative interest, as they are specifically oriented to the vast class of application domains that are describable by means of taxonomic organizations of complex objects. Although the field of TLs has lately been an active area of investigation, only few researchers have addressed the problem of extending these logics with the ability to perform default reasoning. Such extensions would prove of paramount applicative value, as many application domains may be formalized by means of monotonic. TLs only at the price of oversimplification. In this paper we show how we can effectively integrate terminological reasoning and default reasoning, yielding a terminological default logic. The kind of default reasoning we embed in our TL is reminiscent of Reiter's default logic, but overcomes some of its drawbacks by subscribing to the ``implicit'' handling of exceptions typical of the multiple inheritance networks with exceptions proposed by Touretzky and others.