Abstract: An adaptive data compression and decompression system having a fixed dictionary size is disclosed. The compression system builds a data dictionary tree such that each path in the tree represents a string .omega., which was, along with its prefixes, in the input character stream being compressed. The root node and all nodes that do not represent a single character word in the input stream are tagged. Additionally, each non-root node is associated with a unique output code. As strings from the input stream of characters are matched against the tree, the node tags are updated. When a match fails, a new node is added. The new node is associated with a new code and a tag. When the dictionary is full, the root tag identifies the tag of a node that can be deleted. The decompression system builds a similar tapped decompression dictionary.