Studied the importance of retroactive interference (RI) in memory for spatial locations by using a 12-arm radial maze and a standard RI paradigm. 26 male albino Sprague-Dawley rats in the RI condition first learned to choose 4 of the 12 arms, followed by training to a 2nd set of 4 arms. In the control condition for interference, Ss learned the 1st set of arms but were not trained to approach the 2nd set. Thereafter, Ss in each interference condition were assigned to groups (hippocampal, cortical control, or unoperated control), the operations were carried out, and then all Ss were tested for retention of the set of arms learned first. Contrary to predictions of the cognitive map theory (J. O'Keefe and L. Nadel, 1978), RI was found in control Ss. The severe memory deficit found in hippocampals was not influenced by the interference variable. In addition to impaired performance early in relearning, Ss with hippocampal lesions continued to make many errors throughout the 10 wks of testing, including choices to unbaited arms and repeated entries into baited arms. However, hippocampals eventually learned not to reenter unbaited arms. Data indicate a deficit in the selection and utilization of sets of responses and are interpreted as implicating the hippocampus in retrieval processes. (42 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved)