In a data distribution system, data is divided into a number of application data units. A sequence of keys is generated systematically, and a different key is used to encrypt each data unit at the source. At the receivers, corresponding keys are generated and used to decrypt the data units to gain access to the data. The constructions used to generate the keys are such that an intrinsically limited subset of the entire sequence of keys is made available to the user by communicating a selected combination of one or more seed values.