With regard to the deficits in emotion understanding, previous studies have obtained inconclusive and mixed findings. Moreover, most previous studies focus on face emotion recognition. They commonly use isolated facial expressions and ask participants to recognize the emotion. Nevertheless, in real-world interactions, individuals are unlikely to encounter a face without any social context and to be asked directly which emotion does someone feel. Actually, implicit processing of emotions is an important component of social interaction. Therefore, this study aims to use materials of continuously developing emotional situations with higher ecological validity and both implicit and explicit measurements to investigate the emotion understanding ability of autistic children. Twenty-one autistic children and 21 IQ-matched typically developing children (TDC) participated the experiment. In explicit task, materials were emotional situations selected from cartoons which children were familiar with. Each situation contained a sequence of 4 pictures describing one character and a simple situation. According to 20 undergraduates' rating of valence and arousal, these emotional situations were divided into 3 categories of valence: positive, negative, and neutral. In implicit task, the 4th picture of each sequence was replaced by a new picture of the same character but different situation, creating an inconsistent picture sequence. Materials in implicit task contained these inconsistent picture sequences as well as the original consistent sequences. Two groups of children completed both explicit and implicit tasks with 1 week apart and the order of tasks counter-balanced. In explicit task (emotion judgment), each participant was presented with consistent emotional situations randomly and asked to judge whether the protagonists' emotion was happy, unhappy or he don't know. In implicit task (consistency judgment), participants were presented with emotional situations randomly (half were inconsistent) and asked to judge whether the 4 pictures described the same situation. Results showed that: 1. In explicit task, the accuracy of emotion judgment in positive and negative situations for autistic children was significantly lower than that for TDC. Especially in negative situations, there was a significant deficit in emotion understanding for autistic children; 2. In implicit tasks, for both groups, the accuracy of consistency judgment in positive situations was significantly higher than that of neutral and negative situations; 3. The emotion understanding effects were calculated by subtracting the accuracy of neutral situation from that of positive or negative situations. The effect of positive situations for both groups and that of negative situations for TDC became weaker in implicit task. However, the effect of negative situations for autistic children wasn't different between explicit and implicit tasks (close to 0). To conclude, the findings of this study indicate that there are deficits in understanding negative emotion in both explicit and implicit tasks for autistic children. The present study provide new evidence and a new methodological perspective for exploring the emotion understanding among children with autism.