The present study reviews our experience of acute hematogenous osteomyelitis in 450 children over a period of 20 years from 1985 to 2004 at the Nouméa Territorial Hospital in New Caledonia. The objective was to formulate a new theory of the pathogenesis of this affection and to report our research on the disparity in the pathology between temperate countries and our own tropical Pacific area. Only children with an initially normal X-ray and showing symptoms for less than one week were included in the study. Subacute osteomyelitis, infant osteoarthritis and spinal and sacroiliac joint infections were all excluded. All children were treated according to a preestablished protocol including: clinical examination; blood tests; ultrasound, to determine the presence and size of the periosteal elevation and to exclude soft tissue abscess and frequent pyomyositis. Ultrasound was used in the decision to treat with antibiotics alone or with surgery. Computed Tomography was used for deep structures assessment and medical therapy guidance Surgery was limited to open drainage of the subperiosteal abscess only. Regular follow-up of outpatients was continued until normal blood test and X-ray results were achieved. Four hundred and fifty children with a diagnosis of acute hematogenous osteomyelitis were identified, giving an average incidence of 22 new cases per year (range, 12-35). This incidence was two to five times as high as found in Europe. Fifty-three percent of our cases required surgical drainage (vs. 20 % in Europe). Ethnically, 60 % of the children were Melanesian and 20 % Polynesian (both represented less than 50 % of the local population). A similar incidence, about four times as high as in the population of European descent, was reported in Polynesians by our neighbors in New Zealand. The limbs were affected in 90 % of cases, and specifically lower limbs in 70 %. Multiple osseous lesions and systemic infection were recorded in 43 children (9.5 %). Blood cultures and surgical samples were positive in 80 % of cases, and otherwise negative. All the children were successfully treated, without chronic evolution or sequelae needing secondary surgery. The predominant microorganisms isolated were Staphylococcus aureus, in 81 % of cases, none of which were methicillin-resistant, and group A Streptococcus in 7.5 % of cases. A previous study of soft-tissue S. aureus infection showed the presence of Panton-Valentine Leukocidin (PVL) genes in 89 % of cases. These very infrequent genes are responsible for leukotoxic apoptosis, producing leukocidin, causing local acute aggressiveness. A parallel study, in progress for more than a year, is focusing on detecting PVL genes in S. aureus isolated from acute osteomyelitis: in the first nine children analyzed, PVL genes were likewise detected in 89 % of the S. aureus isolated, with no methicillin resistance. Ultrasonography allowed positive diagnosis in 64 % of cases on the day of admission and 84 % by the second day. Because of this very early presence of subperiosteal abscess at the beginning of the disease, and several other issues raised in the present study, we believe that Trueta's theory of acute osteomyelitis pathogenesis does not provide any logical explanation for our anatomoclinical observations. We believe that the primary focus of infection is in the osteoperiosteal area rather than under the growth plate in the metaphyseal bone. The term of Acute Osteo-Periostitis would therefore be much more suitable. A history of blunt trauma was found in 63 % of cases in the present series, and often reported in the literature. We speculate that two forms of infection fixation may develop: a local form, where bacteria carried by the blood stream reach a subperiosteal edema or hematoma secondary to blunt trauma, which is in our opinion the most frequent cause; and a general form, where fixation occurs as single or multifocal osteoperiostitis, and multivisceral locations in severe forms of septicemia. The disparity in this pathology between temperate countries and our own tropical Pacific area is certainly due to PVL-positive S. aureus and ethnic factors. The high prevalence of Melanesian and Polynesian patients confirms that they are at high risk of musculoskeletal infection in New Caledonia as in other Pacific countries, and it is possible that these ethnic groups are genetically susceptible to PVL-positive strains. Level IV. Retrospective case series.