This research identifies a new source of failure to make accurate affective predictions or to make experientially optimal choices. When people make predictions or choices, they are often in the joint evaluation (JE) mode; when people actually experience an event, they are often in the single evaluation (SE) mode. The "utility function" of an attribute can vary systematically between SE and JE. When people in JE make predictions or choices for events to be experienced in SE, they often resort to their JE preferences rather than their SE preferences and overpredict the difference that different values of an attribute (e.g., different salaries) will make to their happiness in SE. This overprediction is referred to as the distinction bias. The present research also specifies when the distinction bias occurs and when it does not. This research contributes to literatures on experienced utility, affective forecasting, and happiness.