Two experiments explored limited capacity inhibitory selective attention processes in working memory. Experiment 1 used a modified Sternberg-type ‘short-term memory scanning’ task, where both irrelevant and relevant memory-set words were included to see if an inhibitory fan effect operated on lexical associates of the should-be-ignored (irrelevant) words. Experiment 2 used a ‘negative priming’ task, where a target letter to be named was flanked by one, two, or three distractor letters to see if an inhibitory fan effect operated on the should-be-ignored letters. Results from both experiments supported the existence of a limited capacity spreading inhibition counterpart to spreading activation. The findings were discussed in terms of a model recently proposed by Neumann and DeSchepper (1991; 1992) in which two selective attention subprocesses (one excitatory and one inhibitory) in the brain each maximise opposed functions within their respective resource limitations in working memory.