PURPOSES: To review in the literature, all the epidemiological, clinical, radiological, histological and therapeutic data regarding chordomas as well as various notochordal entities: ecchordosis physaliphora, intradural and intraparenchymatous chordomas, benign notochordal cell tumors, parachordomas and extra-axial chordomas. To identify different types of chordomas, including familial forms, associations with tuberous sclerosis, Ollier's disease and Maffucci's syndrome, forms with metastasis and seeding. To assess the recent data regarding molecular biology and progress in targeted therapy. To compare the different types of radiotherapy, especially protontherapy and their therapeutic effects. To review the largest series of chordomas in their different localizations (skull base, sacrum and mobile spine) from the literature. MATERIALS: The series of 136 chordomas treated and followed up over 20 years (1972-2012) in the department of neurosurgery at Lariboisière hospital is reviewed. It includes: 58 chordomas of the skull base, 47 of the craniocervical junction, 23 of the cervical spine and 8 from the lombosacral region. Similarly, 31 chordomas in children (less than 18 years of age), observed in the departments of neurosurgery of les Enfants-Malades and Lariboisière hospitals, are presented. They were observed between 1976 and 2010 and were located intracranially (n=22 including 13 with cervical extension), 4 at the craniocervical junction level and 5 in the cervical spine. METHODS: In the entire Lariboisière series and in the different groups of localization, different parameters were analyzed: the delay of diagnosis, of follow-up, of occurrence of metastasis, recurrence and death, the number of primary patients and patients referred to us after progression or recurrence and the number of deaths, recurrences and metastases. The influence of the quality of resection (total, subtotal and partial) on the prognosis is also presented. Kaplan-Meier actuarial curves of overall survival and disease free survival were performed in the entire series, including the different groups of localization based on the following 4 parameters: age, primary and secondary patients, quality of resection and protontherapy. In the pediatric series, a similar analysis was carried-out but was limited by the small number of patients in the subgroups. RESULTS: In the Lariboisière series, the mean delay of diagnosis is 10 months and the mean follow-up is 80 months in each group. The delay before recurrence, metastasis and death is always better for the skull base chordomas and worse for those of the craniocervical junction, which have similar results to those of the cervical spine. Similar figures were observed as regards the number of deaths, metastases and recurrences. Quality of resection is the major factor of prognosis with 20.5 % of deaths and 28 % of recurrences after total resection as compared to 52.5 % and 47.5 % after subtotal resection. This is still more obvious in the group of skull base chordomas. Adding protontherapy to a total resection can still improve the results but there is no change after subtotal resection. The actuarial curve of overall survival shows a clear cut in the slope with some chordomas having a fast evolution towards recurrence and death in less than 4 years and others having a long survival of sometimes more than 20 years. Also, age has no influence on the prognosis. In primary patients, disease free survival is better than in secondary patients but not in overall survival. Protontherapy only improves the overall survival in the entire series and in the skull base group. Total resection improves both the overall and disease free survival in each group. Finally, the adjunct of protontherapy after total resection is clearly demonstrated. In the pediatric series, the median follow-up is 5.7 years. Overall survival and disease free survival are respectively 63 % and 54.3 %. Factors of prognosis are the histological type (atypical forms), localization (worse for the cervical spine and better for the clivus) and again it will depend on the quality of resection. CONCLUSIONS: Many different pathologies derived from the notochord can be observed: some are remnants, some may be precursors of chordomas and some have similar features but are probably not genuine chordomas. To-day, immuno-histological studies should permit to differentiate them from real chordomas. Improving knowledge of molecular biology raises hopes for complementary treatments but to date the quality of surgical resection is still the main factor of prognosis. Complementary protontherapy seems useful, especially in skull base chordomas, which have better overall results than those of the craniocervical junction and of the cervical spine. However, we are still lacking an intrinsic marker of evolution to differentiate the slow growing chordomas with an indolent evolution from aggressive types leading rapidly to recurrence and death on which more aggressive treatments should be applied.