Autism spectrum disorder (ASD) has become widespread neurodevelopmental disorder, which currently can be treated with only few therapeutic options. Furthermore, their effectiveness is limited therefore novel treatment strategies for ASD are needed. This review seeks to address this need by discussing a ketogenic diet (KD) in the context of ASD therapy. KD effects have been examined in animal and human studies. They indicate effectiveness of KD by improving autistic features. Moreover, animal studies have revealed clinically useful information about caloric restriction component of KD, which is not necessary to achieve therapeutic effects. Significantly administration of KD but not β-hydroxybutyrate or acetone has a therapeutic effect on social interactions. Human studies are scarce, however previous researches imply KD as an effective treatment at least in certain types of autism. KD in an altered form as: modified Atkins diet (MAD), ketogenic gluten-free diet with supplemental medium-chain triglyceride (MCT), and John Radcliffe ketogenic diet is an alternative to classic KD. These variants provide better quality of nutrition and are less strict, thus less difficult to maintain. KD is described as safe with limited, easily manageable adverse effects. Taken together human and animal studies would seem to suggest that KD will become part of ASD treatment. However, in order to determine accurate recommendations for all ASD patients, further studies are required.