Autism spectrum disorder (ASD) is a group of developmental diseases characterized by social dysfunction and repetitive stereotype behaviors. Besides genetic mutations, environmental factors play important roles in the development of ASD. Valproic acid (VPA) is widely used for modeling environmental factor induced ASD in rodents. However, traditional VPA modeling is low-in-efficiency and the phenotypes often vary among different batches of experiments. To optimize this ASD-modeling method, we tested “two-hit” hypothesis by single or double exposure of VPA and poly:IC at the critical time points of embryonic and postnatal stage. The autistic-like behaviors of mice treated with two-hit schemes (embryonic VPA plus postnatal poly:IC, embryonic poly:IC plus postnatal VPA, embryonic VPA plus poly: IC, or postnatal VPA plus poly:IC) were compared with mice treated with traditional VPA protocol. The results showed that all single-hit and two-hit schemes produced core ASD phenotypes as VPA single treatment did. Only one group, namely, mice double-hit by VPA and poly:IC simultaneously at E12.5 showed severe impairment of social preference, social interaction and ultrasonic communication, as well as significant increase of grooming activity and anxiety-like behaviors, in comparation with mice treated with the traditional VPA protocol. These data demonstrated that embryonic two-hit of VPA and poly:IC is more efficient in producing ASD phenotypes in mice than the single-hit of VPA, indicating this two-hit scheme could be utilized for modeling environmental factors induced ASD.