Anderson & Green (2001) have recently shown that using an adaptation of the go-no go task, participants can voluntarily inhibit the retrieval of specific memories. We present three experiments in which we try to determine the degree of automaticity involved, and the role of the previous prime-target relation on the development of this inhibitory process. In the first two experiments we manipulated stimulus onset asynchrony at test (100 vs. 700), and the type of pre-experimental cue-target relatedness at study (no relation vs. preexisting). Additionally, we carried out an independent probe test in the three experiments. The results show that inhibitory control is only achieved strategically, it is directly linked to the trained stimulus, and it is obtained with episodically and associatively related pairs of stimuli. We discuss these results in terms of clinical and memory research data. (Contains 3 tables and 1 footnote.)