Tomatoes (Lycopersicon esculentum Mill. cv. Jet Star) were grown under field conditions in open-top chambers at Beltsville, MD. Sets of chambers circulating carbon-filtered air to remove the O3 and non-filtered air with and without an admixture of SO2 at 1256 µg/m3 comprised the treatments. Mean reflectances and their corresponding standard deviations were calculated at 475, 550, 660, and 725 nm for leaflets selected from all chambers. Results of these calculations showed that the reflectance at 550 nm was the most responsive to treatment, while 725 nm showed intermediate response. Since both 475 and 660 nm were shown to be treatment insensitive, each was selected as a basis from which a threshold reflectance level could be established for the assessment of incipient injury. The data revealed that the effects of pollutants on leaflets could be assessed by differencing one of the treatment-sensitive wavelengths, 550 or 725 nm, with respect to one of the treatment-insensitive wavelengths, 475 or 660 nm, and comparing the resulting differences with the appropriate threshold reflectance viz: (R550-Ri)j > (< R550-Ri >)c + 2δic, where i = 475 or 660 nm, j refers to treatments, and (< R550-Ri >)c and δic refer to the average reflectance at 550 nm relative to the reference wavelength i for the controls and standard deviation at the reference wavelength, respectively. The standard deviation is positive because in no instance was (R550-Ri)j less than (< R550-Ri >)c. In combination, these factors functioned as a threshold reflectance. For i = 475 nm, the criterion was more sensitive to abaxial injury, whereas for i = 660 nm, it was found more effective in predicting adaxial effects. To evaluate the criterion, percentages of injured leaves were compared with fruit yield (weight) information. Using 660 nm for the reference wavelength, the percentages of injured leaflets were 38.4, 32.0, and 30.9% for O3 + SO2), O3, and SO2, while the respective yields were 68, 82, and 84% of the control. Using 475 nm, the values for the same sequence were 57.7, 55.1, and 41.3%, demonstrating that both reference wavelengths correctly ordered injured leaflets with respect to yield data.