Objective This study aims to compare the influence of immediate and postponed responses in the detection of prospective memory (PM) target cues and consequent PM retrieval, while also varying the ongoing task demands. This comparison can have important implications in the design of PM tasks, especially because there has been an interchangeable use of both responses without taking into consideration that they might require different mechanisms. Method A total of 32 participants performed a task switching paradigm with an embedded PM task, following a within‐subjects 3 (type of response: no PM response, immediate, postponed) × 3 (switching load: pure, repetition, alternation) design. Results The results yielded no relevant effects of type of response, immediate, or postponed, neither in the PM accuracy nor in the ongoing task performance. However, a significant PM interference effect was found with slower response times in the ongoing activity with PM requirements in comparison with the same task with no PM instructions. Conclusions Overall, given the experimental parameters used, this study supports no behavioral differences between immediate and postponed responses even when the ongoing task is also characterised by different levels of demand.