AIMS: To test the hypothesis that the abnormal fibrous stroma of bone marrow found after treatment of disseminated neuroblastoma represents traumatic scarring. METHODS: Twenty six restaging bone marrow biopsy specimens from 14 children with disseminated neuroblastoma were compared with 37 from 37 children with acute lymphoblastic leukaemia. They were assessed independently by two observers for the presence of abnormal fibrous stroma. All biopsies were performed 23 to 50 days after the initial diagnostic biopsy, and from the same iliac crest. A further nine restaging bone marrow biopsy specimens from previously unbiopsied iliac crests from four children with neuroblastoma were examined and the appearances of the diagnostic bone marrow specimens in those with neuroblastoma recorded. Between diagnostic and restaging biopsies, the children received combination chemotherapy regimens. RESULTS: In neuroblastoma abnormal stromal material was observed in 17 of 26 rebiopsied iliac crests as opposed to only 9 of 37 of those with acute lymphoblastic leukaemia (p < 0.005). In addition, four of nine restaging biopsy specimens taken from previously unbiopsied crests in patients with neuroblastoma contained abnormal stroma. All 11 children with neuroblastoma whose restaging specimens contained abnormal stroma had similar abnormalities in one or more biopsy specimens at diagnosis. CONCLUSIONS: Abnormal fibrous stromal material in restaging marrow biopsy specimens of previously biopsied iliac crests occurs much more frequently in neuroblastoma than in acute lymphoblastic leukaemia. We suggest that factors other than traumatic scarring are responsible. This is supported by finding similar abnormalities in previously unbiopsied iliac crests and in the diagnostic biopsy specimens of children with neuroblastoma.