Evidence suggests that individuals with schizophrenia (SZ) report anticipatory pleasure deficits compared to controls and that these deficits are linked to decreased motivation to engage socially. However, these deficits have been identified via self-report measures of hypothetical pleasant stimuli, leaving it unclear whether they exist in reference to actual social situations. To address this issue, we created a live social interaction that minimized the reliance of higher-order cognitive processes. SZ and control participants were told that they would be playing an "enjoyable sharing game" with another study participant (who was actually a confederate) that involved asking and answering questions (36 interpersonal closeness generation questions; Aron et al., 1997). Participants then reported their current mood and the emotions they anticipated experiencing during the pleasant social interaction. Immediately following the interaction, they reported their experienced emotions. We found that the SZ group anticipated more negative emotion (d = 1.0), but were less accurate in forecasting negative emotion (d = .81), than controls, and these effects were large. There were small, non-significant group differences in anticipation, experience, and accuracy in forecasting of positive emotion (all ds < .29). Also, social anhedonia was positively correlated with anticipated negative affect and negatively associated with experienced positive emotion. At the same time, controls reported finding the interaction to be a more positive emotional experience overall, d = 0.75. This is the first study to show that "anticipatory pleasure deficits" in SZ might actually be heightened anticipated negative emotion and that inaccurate forecasting could be linked to decreased social motivation.