Proactive interference (PI) refers to the finding that memory for recently studied (target) material can be impaired by the prior study of other (nontarget) material. Previous accounts of PI differed in whether they attributed PI to impaired retrieval or impaired encoding. Here, we suggest an integrated encoding-retrieval account, which assigns a role for each of the 2 types of processes in buildup of PI. Employing a typical PI task, we examined (a) the role of encoding processes in PI by recording scalp EEG during study of nontarget and target lists, and (b) the role of retrieval processes in PI by measuring recall totals and response latencies in target list recall. In addition, we measured subjects' working memory capacity (WMC). Behaviorally, the PI effect arose in both recall totals and response latencies, indicating PI at the sampling and the recovery stage of recall. Neurally, we found an increase in electrophysiological activities in the theta frequency band (5-8 Hz) from nontarget to target list encoding, indicating an increase in memory load during target list encoding. The results demonstrate that impaired retrieval and impaired encoding can contribute to PI. They also show that WMC affects PI. For both encoding and retrieval processes, PI was reduced in high-WMC subjects, suggesting that these subjects are able to separate target from nontarget information and create stronger focus on the target material.