Working memory (WM) is the ability to temporarily store and retrieve a limited amount of information during complex cognitive activities, especially in the face of distraction. The dual-component model describes WM as including active maintenance in primary memory (PM) and cue-dependent search and retrieval from secondary memory (SM). Previously, researchers have found that WM training (WMT) fails to enhance SM capacity, a component that mediates the relationship between WM and fluid reasoning (gF). Thus, a double-blinded, placebo-controlled, randomized trial was conducted to elucidate whether retrieval from SM could be targeted using a two-component WMT regimen versus two control conditions: adaptive one-component WMT targeting solely PM capacity and non-adaptive one-component WMT. Participants were 174 adolescents, aged 10 to 13 years, who were assessed before, after, and six-moths following training. Retrieval from SM was measured using delayed free recall tasks, far transfer to gF was assessed with matrix reasoning and verbal inference tests, and far transfer to academic performance was assessed with reading and math tests. It was predicted that solely two-component WMT would enhance retrieval from SM and result in far transfer. ANCOVAs with pre-test scores as the covariate indicated that two-component participants increased total errors over controls. There were no significant differences between the groups on recall latency, total correct, or gF measures. The non-adaptive one-component group significantly improved on reading, although a drop in the other two groups drove the effect. Additional research is needed to elucidate whether theoretically-motivated WMT can positively impact higher-level cognition through SM retrieval mechanisms. [This paper was published in "Journal of Cognitive Enhancement" v1 p455-477 2017.]