The early start Denver model (ESDM) has been extensively studied as a promising early intervention approach for young children with autism spectrum disorder (ASD). Various methodological drawbacks from earlier ESDM investigations must be rectified to expand the application scopes. For this purpose, the present study recruited a very large sample of 249 autistic children (aged 24-47 months), and used a randomized controlled design to compare outcomes from a mixed ESDM (M-ESDM) intervention with a mixed discrete trial teaching (M-DTT) intervention which remains one of the most commonly-used programming for early intervention. Over the course of a 12-week period, both groups (i.e., M-ESDM and M-DTT groups) received 25 h of intensive intervention per week using individual, group, and parent coaching techniques. Findings showed that: (i) the M-ESDM significantly outperformed the M-DTT in enhancing children's developmental abilities in gross motor and personal-social skills for toddlers and preschoolers, as well as in language for preschoolers with mild/moderate ASD and toddlers; and (ii) the M-ESDM dramatically reduced the severity of autistic symptoms in toddlers with severe ASD only, when compared to the M-DTT. However, the M-ESDM did not outperform the M-DTT in terms of improving children's developmental abilities in adaptability and fine motor for toddlers and preschoolers, as well as in language for preschoolers with severe ASD. In addition, when compared to the M-DTT, the M-ESDM did not show an advantage in reducing the severity of autistic symptoms in toddlers with mild/moderate ASD and preschoolers. Clinical Trial Registration: Chinese Clinical Trial Registry. Registration number ChiCTR200039492.