Employing representation understanding task and rule learning task to investigate the effect of parasocial interaction on 2-5 years old children’s learning from video, young children’s performances were compared among three conditions: social interaction live condition, parasocial interaction video condition and non-parasocial interaction video condition. Results from the representation understanding task indicated that a video deficit phenomenon occurred in 2 year old children’s learning from video; After age 2.5, especially for 3-4 years old children, the performance under parasocial interaction video condition was the same as under social interaction live condition，it implied that parasocial interaction enhanced young children’s participation, and effectively improved children's attention to the video information, thus it contributed to young children's use of video information to solve problems. Under the non-parasocial interaction video condition, the performance of 3 to 4 years old age groups not only did not rise, but declined compared to the performance of 2.5 years old age group, and the performance were significantly lower than the performance under parasocial interaction video condition. Children after 3 years of age, along with the development of self-awareness and self-esteem, their participation and learning might be inhibited under the non-parasocial interaction video condition. Results from the rule learning task indicated that for 2.5 years old age groups, performances were poor under all three conditions due to the limitation of children's cognitive development. For children aged 3 to 4 years old, their performances steadily increased under parasocial interactive video condition and social interaction live condition, and were significantly high compared with non-parasocial interaction video condition. It implied that children’s ability to understand the rules developed rapidly, and learning under parasocial interactive video condition were better than under non-parasocial interaction video condition. For children age 4.5 and above, performances under non-parasocial interactive video condition were greatly improved, because children’s ability of attention and initiative to learning developed rapidly. The prominent role of parasocial interaction in video learning therefore was reduced. Based on the results, the authors concluded that age from 2.5 to 4.5 is the best period for effective parasocial interactive video learning. The authors put forward a suggestion about producing video programs for 2 to 5 years old children that parasocial interaction cues should be considered appropriately in the programs.