Prospective memory (PM) refers to the ability to remember to realize a delayed intention at a future time. The cognitive processes underlying PM have attracted a lot of research attentions. According to the Multiprocess Theory, PM can be realized through either automatic or strategic monitoring processes depending on a variety of factors. Focality of PM cues represents a critical factor that influences the type of processing recruited for PM retrieval. Specifically, the retrieval of intention occurs more spontaneously in focal PM tasks, whereas it relies more on strategic monitoring in nonfocal PM tasks. However, most of the previous studies drew conclusions based on PM cost (the longer reaction time of ongoing task when a PM task was embedded than ongoing task alone), the performance of PM trials was under represented. In the present study, we tested the Multiprocess Theory by embedding the PM task in a visual search task through the eye-tracking technique, we analyzed both the ongoing trials (representing PM monitoring) and PM trials (representing PM retrieval). Experiment One recruited 28 participants and their eye movements were recorded during the focal and nonfocal PM tasks. While performing the target searching task, participants have to remember to execute an intention (press the “space key”) based on either two specific words (each one occurred six times, this was the focal PM task) or twelve animal words (each cue occurred only once, this was the nonfocal PM task). For PM trials, the time from first fixation to response of PM cue served as an index of PM retrieval; while for ongoing trials, the total fixation counts of distracters reflected strategic monitoring. The repetition times of PM cue words were different between focal and nonfocal conditions in this experiment, so we conducted Experiment Two to control for potential confounding caused by this problem. In Experiment Two, all PM cues repeated the same times. Results of Experiment One indicated that the time to first fixation of PM cue (t(27) = -3.57, p = 0.006, Cohen’s d = -0.74) and the time from first fixation to response of PM cue (t(27) = -2.95, p = 0.001, Cohen’s d = -0.70) in the focal task were shorter relative to those in the nonfocal task, and the total fixation counts of distracters in the nonfocal task were more than those in the focal task (t(27) = 2.95, p = 0.007, Cohen’s d = 0.55). There was an increase in total fixation durations from distractors, M = 267.71 ms, to target hits, M = 305.87 ms, to PM hits, M = 407.81 ms, in the focal task. For the nonfocal task, there was no significant difference in total fixation durations between target hit (M = 302.01 ms) and distracters (M = 295.37 ms, p = 0.856). These results were also observed in Experiment Two, which further demonstrated the different underlying processing mechanisms between focal and nonfocal PM. Our study revealed that the focal and nonfocal PM involved different processes related to the monitoring and retrieval of PM intentions, which supported the Multiprocess Theory.