SETTING: Metropolitan areas of England, including London boroughs, in 1991. OBJECTIVE: To investigate the relative importance of deprivation, immigration and the elderly in explaining variations in tuberculosis rate. DESIGN: A retrospective study using multiple Poisson regression models to assess the interrelationship between various population parameters. RESULT: Significant differences ere observed between London and other metropolitan districts in the measures of tuberculosis, immigration and the elderly. In addition, all population parameters were significantly intercorrelated in London: areas with a high proportion of immigrants had high levels of deprivation and low proportions of elderly. In other metropolitan districts, only immigration and the Jarman index were significantly associated, and removing the immigration component from the index removed this statistical significance. Multiple Poisson regression models revealed that the immigrant index had the strongest explanatory power in explaining tuberculosis rates, but there were significant interactions between this and measures of urban deprivation indices. That is, there was a greater effect of increasing deprivation at lower levels of immigration than at higher levels. This phenomenon was more pronounced in London boroughs than other metropolitan districts. The elderly index had no significant influence on tuberculosis rates. CONCLUSION: Although the association between tuberculosis and deprivation previously reported for the city of Liverpool is confirmed across all urban areas of England, the immigrant proportion of the population has a greater statistical power in explaining variations in rates of urban tuberculosis. However, tuberculosis notifications can be most accurately predicted by combining both measures than by either one alone.