Prospective memory (PM) supports the planning and execution of future activities, and is particularly important in applied settings. We investigate a new response method that aims to improve PM accuracy by integrating the responses to an occasional PM task and a routine ongoing lexical-decision task. Instead of the most common three-choice method where the PM response replaces the ongoing response, participants were obligated to make explicit PM (present vs. absent) and ongoing (word vs. non-word) classifications on every trial through a four-choice response. Although replacement and obligatory responses were initially similar in PM accuracy, an advantage emerged with practice for the new obligatory method that was not simply due to slower responding associated with making four versus three choices. The nature of the errors differed between methods, with obligatory responding being characterised by fast PM errors and replacement by slower errors, suggesting avenues for further potential improvements in PM accuracy.