Abstract Background The term congenital ocular motor apraxia (COMA), coined by Cogan in 1952, designates the incapacity to initiate voluntary eye movements performing rapid gaze shift, so called saccades. While regarded as a nosological entity by some authors, there is growing evidence that COMA designates merely a neurological symptom with etiologic heterogeneity. In 2016, we reported an observational study in a cohort of 21 patients diagnosed as having COMA. Thorough re-evaluation of the neuroimaging features of these 21 subjects revealed a previously not recognized molar tooth sign (MTS) in 11 of them, thus leading to a diagnostic reassignment as Joubert syndrome (JBTS). Specific MRI features in two further individuals indicated a Poretti–Boltshauser syndrome (PTBHS) and a tubulinopathy. In eight patients, a more precise diagnosis was not achieved. We pursued this cohort aiming at clarification of the definite genetic basis of COMA in each patient. Results Using a candidate gene approach, molecular genetic panels or exome sequencing, we detected causative molecular genetic variants in 17 of 21 patients with COMA. In nine of those 11 subjects diagnosed with JBTS due to newly recognized MTS on neuroimaging, we found pathogenic mutations in five different genes known to be associated with JBTS, including KIAA0586, NPHP1, CC2D2A, MKS1, and TMEM67. In two individuals without MTS on MRI, pathogenic variants were detected in NPHP1 and KIAA0586, arriving at a diagnosis of JBTS type 4 and 23, respectively. Three patients carried heterozygous truncating variants in SUFU, representing the first description of a newly identified forme fruste of JBTS. The clinical diagnoses of PTBHS and tubulinopathy were confirmed by detection of causative variants in LAMA1 and TUBA1A, respectively. In one patient with normal MRI, biallelic pathogenic variants in ATM indicated variant ataxia telangiectasia. Exome sequencing failed to reveal causative genetic variants in the remaining four subjects, two of them with clear MTS on MRI. Conclusions Our findings indicate marked etiologic heterogeneity in COMA with detection of causative mutations in 81% (17/21) in our cohort and nine different genes being affected, mostly genes associated with JBTS. We provide a diagnostic algorithm for COMA.