This paper discusses the issue of exposure measurement error in time-series studies of air pollution and health. Exposure misclassification is a well-known limitation in epidemiological studies. For many pollutants, exposure occurs over time and in multiple locations, making accurate estimation of individual exposure difficult. Researchers have addressed this by limiting error through study design, using nested validation studies to estimate error, and adjusting for measurement error in statistical analyses. The paper provides an overview of measurement errors in linear regression, distinguishing between Berkson and classical error types, and univariate and multivariate predictor cases. It proposes a conceptual framework for evaluating measurement errors in log-linear regression used for time-series studies of particulate air pollution and mortality. The paper also presents new analyses of data on particulate matter < 10 μm in aerodynamic diameter from the Particle Total Exposure Assessment Methodology Study. The paper highlights the importance of understanding measurement error in interpreting epidemiological studies on air pollution, particularly time-series analyses. It discusses the consequences of measurement error in regression models, distinguishing between classical and Berkson error models. Under the classical model, measurement error leads to biased estimates of regression coefficients, while under the Berkson model, the estimates are unbiased but have higher variance. The paper also addresses the issue of measurement error in multipollutant models and its implications for estimating the independent effect of a pollutant in a mixture. It discusses the potential for bias in regression coefficients when using ambient concentrations instead of personal exposure data, and presents a method for adjusting for measurement error using regression calibration. The paper concludes that measurement error can significantly affect the interpretation of epidemiological studies on air pollution, particularly time-series analyses. It emphasizes the need for further research to better understand the sources and consequences of measurement error in these studies.This paper discusses the issue of exposure measurement error in time-series studies of air pollution and health. Exposure misclassification is a well-known limitation in epidemiological studies. For many pollutants, exposure occurs over time and in multiple locations, making accurate estimation of individual exposure difficult. Researchers have addressed this by limiting error through study design, using nested validation studies to estimate error, and adjusting for measurement error in statistical analyses. The paper provides an overview of measurement errors in linear regression, distinguishing between Berkson and classical error types, and univariate and multivariate predictor cases. It proposes a conceptual framework for evaluating measurement errors in log-linear regression used for time-series studies of particulate air pollution and mortality. The paper also presents new analyses of data on particulate matter < 10 μm in aerodynamic diameter from the Particle Total Exposure Assessment Methodology Study. The paper highlights the importance of understanding measurement error in interpreting epidemiological studies on air pollution, particularly time-series analyses. It discusses the consequences of measurement error in regression models, distinguishing between classical and Berkson error models. Under the classical model, measurement error leads to biased estimates of regression coefficients, while under the Berkson model, the estimates are unbiased but have higher variance. The paper also addresses the issue of measurement error in multipollutant models and its implications for estimating the independent effect of a pollutant in a mixture. It discusses the potential for bias in regression coefficients when using ambient concentrations instead of personal exposure data, and presents a method for adjusting for measurement error using regression calibration. The paper concludes that measurement error can significantly affect the interpretation of epidemiological studies on air pollution, particularly time-series analyses. It emphasizes the need for further research to better understand the sources and consequences of measurement error in these studies.