Abstract Visibility trends which reflect changes in optical air quality due to air pollution are examined. Two methods of analysis of National Climatic Center visibility data are considered: cumulative percentiles and ridits. Each is applicable to data having nonstandard probability distribution functions. The dependence of trend lines derived from each method of analysis on meteorology is explored by application to mideastern U.S. sites. The most representative data base includes midday observations in the absence of precipitation and nigh relative humidity. A qualitative index of the visibility trend is given by the net percentage change in visibility over the 1948–1978 period obtained from a linear least-squares fit to the trend line. When carefully applied, the 60th cumulative percentile trend line and the mean ridit trend line are in complete concurrence.