The study investigated the effect of selection cues in working memory (WM) on the fate of not-selected contents of WM. Experiments 1A and 1B showed that focusing on 1 cued item in WM does not impair memory for the remaining items. The nonfocused items are maintained in WM even when this is not required by the task. Experiments 2 and 3 showed that items that were once focused in WM remain strengthened after the focus shifts away from them. When defocused items are presented as mismatching recognition probes, they are rejected better than other mismatching probes (Experiments 2 and 3). When a defocused item was later cued again, such that the focus had to shift back to it, that item was recognized better than an item cued for the first time (Experiment 3). The results support the distinction between mechanisms for temporary maintenance and the focus of attention in WM, and they challenge theories that explain maintenance and focusing by the same mechanisms, such as a limited number of slots or a limited resource.