1. The aim of the study was to determine whether geographical variation in the incidence of childhood-onset diabetes was associated with levels of nitrate in drinking water.
2. Population-based registers were the source of cases of childhood diabetes occurring between 1990 and 1996 and nitrate sample data were supplied by the water companies for relevant water supply zones for 1990-1996.
3. Estimates of the population exposure to nitrate were taken as the mean of the monthly mean nitrate levels in 594 water supply zones. Mean nitrate levels were ten times higher in the English study area (22.94 mg l^-1) than Scotland (2.07mg l^-1) with a wider range and more variability. Nitrate levels are dependent on water source and catchment area. The EC maximum acceptable concentration for nitrate in drinking water is 50mg l^-1.
4. An ecological analysis was conducted in the former Oxfordshire Regional and Leicestershire District health authorities, all of Scotland and all areas combined. Factors were examined separately followed by multivariate modelling which accounted for the demographic factors associated with childhood diabetes i.e. socio-economic status, migration and ethnicity.
5. In both the Scottish and English study areas there was no significant geographical variation of diabetes incidence. No significant associations were observed between diabetes incidence and nitrate in drinking water when unadjusted by the demographic factors.
6. Multivariate modelling of the English data, adjusting for all factors, showed no association with childhood diabetes at the highest nitrate levels. There was no evidence of a ‘dose response’ between increasing levels of nitrate and childhood diabetes. However, a significantly raised risk of childhood diabetes was restricted to areas with medium levels of nitrate (16.7-22.2 mg/l). There is no apparent biologically plausible explanation for this observation which may represent a confounding effect.
7. After adjusting for the three demographic variables in Scotland and the larger combined study area data set no significant risks were identified for childhood diabetes and nitrate levels in drinking water.

In conclusion, the overall findings from this ecological analysis do not provide evidence of any consistent or plausible associations between childhood diabetes and levels of nitrate in drinking water in the geographical areas of study.