Existing evidence has linked individual differences in emotion reaction intensity to both enhanced and decreased psychological well-being. We propose that this contradiction is related to methodological shortcomings in some existing research. We present a novel emotion reactivity task capable of addressing these methodological shortcomings by continuously measuring the subjective intensity of individual emotion episodes with high temporal resolution. Four studies were conducted (total n = 499). In Studies 1, 2, and 4, participants continuously reported their emotions while viewing objectively pleasant or unpleasant images. Thousands of reaction intensities were coded using algorithms developed for this purpose. We expected that people showing more intense emotion reactions, regardless of valence, would report greater subjective well-being in the lab and in daily life. One reason that such situationally-congruent reactions might be beneficial is that that they enable more flexible situationally-appropriate behavior. In Study 3, participants were asked to rate their emotional responses to pleasant and unpleasant images. Following this, people choose a location for their Self avatar within a computerized environment that included one image of each valence. We expected that the tendency to report intense emotion responses to these images would predict both adaptive location choice and subjective well-being. Results confirmed most major hypotheses: more intense reactions to both positive and negative stimuli were predictive of greater subjective well-being in the lab and in daily life, and analogous reactivity patterns were associated with more flexible, adaptive avatar placement. The results suggest that a key feature of maladaptive emotion generation systems (and lower well-being) may not be overly intense reactions as has been suggested, but a failure to flexibly adapt emotion output to match changing circumstances.