A hashing data storage and retrieval arrangement whose storage capacity is unaffected by collisions. A first memory serves as a hash index table, for storing pointers at each address location corresponding to a hash value generated by hashing a key data word. Each pointer is the address of a location in a second memory, which has a separate storage location for each key data word, its associated data, and a further pointer which is the address of the next key data word resulting from a collision during hashing. Preferably a pipeline register between the two memories permits hashing of a subsequent key data word while accessing of the second memory is still in progress.