Adolescents with disruptive behavior disorders are reported to show deficits in empathy and emotion recognition. However, prior studies have mainly used questionnaires to measure empathy or experimental paradigms that are lacking in ecological validity. We used an empathic accuracy (EA) task to study EA, emotion recognition, and affective empathy in 77 male adolescents aged 13-18 years: 37 with Conduct Disorder (CD) and 40 typically-developing controls. The CD sample was divided into higher callous-emotional traits (CD/CU+) and lower callous-unemotional traits (CD/CU-) subgroups using a median split. Participants watched films of actors recalling happy, sad, surprised, angry, disgusted or fearful autobiographical experiences and provided continuous ratings of emotional intensity (assessing EA), as well as naming the emotion (recognition) and reporting the emotion they experienced themselves (affective empathy). The CD and typically-developing groups did not significantly differ in EA and there were also no differences between the CD/CU+ and CD/CU- subgroups. Participants with CD were significantly less accurate than controls in recognizing sadness, fear, and disgust, all ps < 0.050, rs ≥ 0.30, whilst the CD/CU- and CD/CU+ subgroups did not differ in emotion recognition. Participants with CD also showed affective empathy deficits for sadness, fear, and disgust relative to controls, all ps < 0.010, rs ≥ 0.33, whereas the CD/CU+ and CD/CU- subgroups did not differ in affective empathy. These results extend prior research by demonstrating affective empathy and emotion recognition deficits in adolescents with CD using a more ecologically-valid task, and challenge the view that affective empathy deficits are specific to CD/CU+.