The ability to generate memory strategies is a key factor in the performance of episodic memory tasks. Whether the ability to generate memory strategies exerts an influence in the performance of memory tests in the elderly population is still a matter of debate. Here we present results from an experimental memory task (Test of Memory Strategies, TMS), comprised of five lists of words starting from an incidental learning task, and four more lists which progressively gain in their external organization of the material, reducing the necessity of mobilizing complex memory strategies. TMS has been applied to four groups of elderly patients (amnestic Mild Cognitive Impairment - aMCI, multidomain (mMCI), and Vascular Cognitive Impairment - VCI and Depression) and a healthy aging group. In conditions with low organization of the material, the mMCI and VCI groups (both showing a greater executive function impairment) have shown lower performance. However, as the material was progressively organized, they improved their performance. The aMCI group showed similar performance to the control group at the lower level of external organization but did not improve performance in conditions with a high level of external organization. The mMCI and VCI groups showed lower performance on all TMS conditions compared with controls. Discriminant analysis revealed 90% sensitivity and specificity to differentiate between groups based on TMS conditions. These results indicate how executive functions influence performance on memory tasks in elderly subjects with different neuropsychological profiles.