S_mmary The West Midlands Regional Children's Tumour Research Group holds high-quality data from 1957 on all childhood cancers in the West Midlands Health Authority region. Since it has been reported that malignant germ cell tumours are increasing in incidence in the north-west of England, we undertook to examine rates in this region and to map the distribution of cases in order to assess any geographical changes in incidence rates. We identified a total of 102 malignant germ cell tumours (MGCTs) between 1957 and 1992.The average age-standardised rate was 1.6 per million per year in the period 1957-74 and 3.6 per million per year during 1975-92, a significant increase (P = 0.0004). Particular increases were noted in older children (10-14 years); P = 0.0002) and in yolk sac (endodermal sinus) tumours (P= 0.004 All cases were subject to pathology review by a panel of three specialist paediatric pathologists in order to verify the diagnosis. In the cases where there was no material available for review, the details were scrutinised by the senior pathologist (AHC) and clinician (JRM) and the diagnosis confirmed or rejected on the available evidence.Annual age-standardised incidence rates (ASRs) were calculated by the direct method (Parkin et al., 1988) using quinquennially age-grouped population figures, derived from mid-year population estimates produced by the West Midlands Regional Health Authority. Changes in the incidence trend of these ASRs were assessed using simple linear regression (Armitage and Berry, 1987) and also a cumulative sum (Cusum) technique (Wetherill, 1977).In addition, in order to assess possible temporal changes in incidence in detail, we chose a priori to divide our 36 year time period into two 18 year groupings (1957-74 and 1975-92).Geographical analysis was carried out based on the postcode of the patient's address at diagnosis, which was then assigned to health district (HD), this being the smallest unit for which consistent population figures were available for the entire time period.The West Midlands County was created in 1974, incorporating Coventry into the existing West Midlands conurbation (Byrne, 1983). Before that year, the boundaries of the WMRHA were coterminous with the shire counties for which population data are available. In order to investigate the incidence in more detail, we examined the rates in two areas of contrasting environments, broadly encompassing the highly populated, industrial areas of the conurbation and the surrounding largely rural areas. (The individual health districts which make up these two regions are shown in Figure