Objectives To study the association between proximity to air polluting industrial facilities and mortality in the Basque Country (Spain) in the 1996–2003 period.

Methods A cross-sectional ecological study with 1465 census sections (CS) as units of analysis with a mean population of 1257 inhabitants. Association of CS mortality with proximity of industries of the European Pollutant Emission Register was studied by type of industrial activity and adjusted for social deprivation. Two distance thresholds (1 km and 2 km) were used as proxies for exposure in a ‘near versus far’ analysis. Causes of mortality studied were: all causes; tracheal, bronchial, and lung cancer; haematological tumours; ischaemic heart disease; cerebrovascular diseases; chronic diseases of the lower respiratory tract; and breast cancer (in women). Poisson's generalised linear mixed models (GLMM) with two random effects (heterogeneity and structured spatial variability) were used in a fully Bayesian environment.

Results Men living in sections within 1 km from energy production industries had greater mortality from tracheal, bronchial, and lung cancer [CI_90% 6% to 53%] as compared with people living further. Women had greater mortality from ischaemic heart disease [CI_90% 1% to 17%] and respiratory illness [CI_90% 1% to 24%] within 2 km from metal-processing industries. On the contrary, within the 1 km buffer from mineral industries, mortality was lower for all causes [CI_90% −20% to −6%] and for ischaemic heart disease [CI_90% −40% to −10%] in women, and from respiratory diseases in men [CI_90% −39% to −4%], while it was greater for breast cancer in women [CI_90% 2% to 28%] within the 2 km buffer.

Conclusions Analysis of mortality by census sections is a helpful exploratory tool for investigating environmental risk factors and directing actions to sites and risk factors with a greater impact on health. Further epidemiological and environmental investigations around metal-processing and energy-producing plants are required.