Objective To examine the association between a biomarker of exposure to secondhand smoke (salivary cotinine concentration) and cognitive impairment. Design Cross sectional analysis of a national population based study. Setting Stratified random sample of households throughout England. Participants 4809 non-smoking adults aged 50 years or more from the 1998,1999, and 2001 waves of the Health Survey for England who also participated in the 2002 wave of the English Longitudinal Study of Ageing and provided saliva samples for cotinine assay and a detailed smoking history. Main outcome measure Cognitive impairment as defined by the lowest 10% of scores on a battery of neuropsychological tests. Results Participants who did not smoke, use nicotine products, or have salivary cotinine concentrations of 14.1 ng/ml or more were divided into four equal size groups on the basis of cotinine concentrations. Compared with the lowest fourth ofcotinine concentration (0.0-0.1 ng/ml) the odds ratios (95% confidence intervals) for cognitive impairment in the second (0.2-0.3 ng/ml), third (0.4-0.7 ng/ml), and highest fourths (0.8-13.5 ng/ml) were 1.08 (0.78 to1.48), 1.13 (0.81to1.56), and1.44 (1.07 to 1.94; P for trend 0.02), after adjustment for a wide range of established risk factors forcognitive impairment. A similar pattern of associations was observed for never smokers and former smokers. Conclusions Exposure to secondhand smoke may be associated with increased odds of cognitive impairment. Prospective nationally representative studies relating biomarkers of exposure to cognitive decline and risk of dementia are needed.