Delayed diagnosis of paediatric cancers is reported regularly and is a source of remorse for physicians and parents and a leading cause of malpractice claims. We did a systematic review of information about the distribution, determinants, and consequences of time to diagnosis of paediatric malignancies and compared these findings with those of court-appointed expert witnesses in malpractice claims in Canada and France. Time to diagnosis varied widely between tumour types in the 98 relevant studies (medians ranged from 2–260 weeks) without any significant decrease with time. Determinants of a long delay in diagnosis included older age, qualification of the first physician contacted, non-specific symptoms, histological type, and tumour localisation. Delayed diagnosis was associated with poor outcome for retinoblastoma and possibly for leukaemia, nephroblastoma, and rhabdomyosarcoma (data were insufficient for definitive conclusions). It was not associated with an adverse outcome for most CNS tumours, osteosarcoma or Ewing’s sarcoma, and, paradoxically, was frequently associated with better outcomes than was short time to diagnosis in these cancers. A third of the court-appointed experts provided testimony concordant with the medical literature. The relations between delay in diagnosis and outcome are complex and probably depend more on tumour biology than on parental or medical factors.