The use of geographic data to interrelate mortality rates and environmental variables is an important approach to determining reasons for the variations in disease frequency. An alternative methodology to standard regression analysis was recently suggested, which is particularly useful when there are a large number of geographic areas and relatively few deaths. An extension of this technique is described, which includes a modified regression approach using areas aggregated according to their level of the environmental factor. An example is given relating nasal cancer mortality to the distribution of industry in England and Wales. The known hazards in the furniture and leather industries are indicated and other potential risk industries, including tailoring, are suggested.