Abstract Empathy is a multidimensional construct that relies on affective and cognitive component processes. A few studies have reported impairments of both cognitive and affective empathy components in patients with schizophrenia. It is, however, not known whether these difficulties are already present at psychosis onset. The affective and cognitive components of empathy were thus assessed in 31 patients with first-episode psychosis (FEP) and 31 matched healthy controls using the Interpersonal Reactivity Index (IRI). Our results were then compared to previous studies of empathy in patients with more chronic schizophrenia via a meta-analysis. In addition, we also assessed the relationship between empathy ratings, Mentalizing performance and clinical symptoms. Contrary to what has been reported in people with more chronic schizophrenia, the IRI ratings did not significantly differ between FEP and controls in our study, though a trend was observed for the Personal distress scale. For the Perspective taking scale, our meta-analysis revealed a significantly lower effect size in this study with FEP patients relative to previous schizophrenia studies. In the FEP group, the IRI ratings were not related to positive, negative or general psychopathology symptoms, but a significant relationship emerged between the Liebowitz Social Anxiety Scale and Perspective taking (negative correlation). In addition, a significant positive correlation was observed between the Empathic concern subscale and our theory of mind task. This study supports the idea that the cognitive component of empathy is less affected in patients with first-episode psychosis relative to patients with more chronic schizophrenia, and the impairments reported in previous reports with more chronic populations should be interpreted in light of a possible deterioration of this cognitive skill. The findings also provide some insight into the relationship between empathy and clinical symptoms such as social anxiety.