Neurodevelopmental disorders (NDDs) constitute a major health issue with >10% of the general worldwide population affected by at least one of these conditions-such as Autism Spectrum Disorders (ASD) and Attention Deficit Hyperactivity Disorders (ADHD). Each NDD is particularly complex to dissect for several reasons, including a high prevalence of comorbidities and a substantial heterogeneity of the clinical presentation. At the genetic level, several thousands of genes have been identified (polygenicity), while a part of them was already involved in other psychiatric conditions (pleiotropy). Given these multiple sources of variance, gathering sufficient data for the proper application and evaluation of machine learning (ML) techniques is essential but challenging. In this chapter, we offer an overview of the ML methods most widely used to tackle NDDs complexity-from stratification techniques to diagnosis prediction. We point out challenges specific to NDDs such as early diagnosis, that can benefit from the recent advances in the ML field. These techniques also have the potential to delineate homogeneous subgroups of patients that would enable a refined understanding of underlying physiopathology. We finally survey a selection of recent papers that we consider as particularly representative of the opportunities offered by contemporary ML techniques applied to large open datasets, or that illustrate the challenges faced by current approaches to be addressed in the near future.