AIM To correlate clinical, radiological, and biochemical features with genetic findings in children with bilateral basal ganglia lesions of unknown aetiology, and propose a diagnostic algorithm for early recognition. METHOD Children with basal ganglia disease were recruited in a 2‐year prospective multicentre study for clinical, biomarker, and genetic studies. Radiological pattern recognition was examined by hierarchical clustering analysis. RESULTS We identified 22 genetic conditions in 30 out of 62 paediatric patients (37 males, 25 females; mean age at onset 2y, SD 3; range 0–10y; mean age at assessment 11y, range 1–25y) through gene panels (n=11), whole‐exome sequencing (n=13), and mitochondrial DNA (mtDNA) sequencing (n=6). Genetic aetiologies included mitochondrial diseases (57%), Aicardi–Goutières syndrome (20%), and monogenic causes of dystonia and/or epilepsy (17%) mimicking Leigh syndrome. Radiological abnormalities included T2‐hyperintense lesions (n=26) and lesions caused by calcium or manganese mineralization (n=9). Three clusters were identified: the pallidal, neostriatal, and striatal, plus the last including mtDNA defects in the oxidative phosphorylation system with prominent brain atrophy. Mitochondrial biomarkers showed poor sensitivity and specificity in children with mitochondrial disease, whereas interferon signature was observed in all patients with patients with Aicardi–Goutières syndrome. INTERPRETATION Combined whole‐exome and mtDNA sequencing allowed the identification of several genetic conditions affecting basal ganglia metabolism. We propose a diagnostic algorithm which prioritizes early use of next‐generation sequencing on the basis of three clusters of basal ganglia lesions.