Research showed that the four Cognitive Interview (CI) mnemonics used individually are unequally effective. We propose to test (i) their benefit when used within the same free recall phase and (ii) an original instruction, 'guided peripheral focus' (GPF). In two studies, 84 and 42 students were interviewed with Structured Interviews (SI), CIs or CI variations about a film viewed 1 week before. Results indicated that (i) if a CI variation with the GPF instead of the 'perspective' elicits more correct information than an SI or a CI, variations replacing the 'perspective' or the 'order' and 'perspective' with control instruction(s) do not; (ii) a partial CI integrating the 'everything', 'context' and GPF increases correct information compared with an SI, whereas the CI does not. We will discuss in which extent these results demonstrate the effectiveness of the 'everything', 'context' and 'GPF' and a potential lack of benefit of the 'order' and 'perspective'.