Attention deficit hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) are highly comorbid neurodevelopmental conditions. There is an ongoing debate regarding the nature of their overlap. Behavioral symptoms and cognitive profiles indicate differences between the conditions, but genetic studies and neuroimaging investigations suggest at least some shared etiology. The current study investigated if functional connectivity can be used to distinguish ADHD and ASD using a machine-learning approach. Towards this aim, we trained a machine learning algorithm to distinguish ASD and ADHD cases from each other and from comparison cases in a total sample of 805 cases, comprising of 243 ASD cases, 164 ADHD cases, and 398 comparison cases between 7 and 21 years of age. We compared the performance of the best performing machine learning algorithm (l2-regularised support vector classification) when classifying unseen cases of ADHD, ASD, and CMP. The results indicated lower classification performance when distinguishing ADHD from ASD compared to classifying diagnostic groups vs a typical comparison group. The model trained to distinguish ASD and comparison cases performed equally well when tasked with classifying ADHD vs CMP. A Bayesian analysis gave strong evidence for similarity ADHD and ASD. The ADHD and ASD group showed overlap in connections of the right ventral attention network, the salience network, and the default mode network. In sum, these results suggest a substantial overlap in functional brain connectivity between ADHD and ASD. We discuss the implications of these findings for the quest to identify functional neuroimaging biomarkers and provide recommendation for future research.