Many researchers report that people have an optimistic bias when making predictions, but sometimes cautious realism is found. One resolution is that future thinking has two steps: The desired outcome is imagined first, followed by a sobering reflection on potential difficulty of getting there. Five experiments supported this two-step model (USA and Norway; N=3,213; 10,433 judgments), showing that intuitive predictions are more optimistic than reflective predictions. Participants were randomly assigned to rely on fast intuition under time-pressure or slow reflection after time-delay. In Experiment 1, participants in both conditions thought positive events were more likely to happen to them than to other people and that negative events were less likely, replicating the classic finding of “unrealistic optimism”. Crucially, this optimistic tendency was significantly stronger in the intuitive condition. Participants in the intuitive condition also relied more on heuristic problem-solving (CRT). Experiments 2-3 found that participants in the intuitive condition thought they were at lower health risk than participants in the reflective condition. Experiment 4 provided a direct replication, with the additional finding that intuitive predictions were more optimistic only for oneself (and not about the average person). Experiment 5 failed to identify any intuitive difference in perceived reasons for success versus failure, but observed intuitive optimism in binary prediction of a future exercise habit. Experiment 5 also found suggestive evidence for a moderating role of social knowledge: Reflective predictions about oneself became more realistic than intuitive predictions only when the person’s base-rate beliefs about other people were fairly accurate.