Differential Optical Absorption Spectroscopy (DOAS) is commonly used to measure gas emissions from volcanoes. DOAS instruments measure the absorption of solar ultraviolet (UV) radiation scattered in the atmosphere by sulfur dioxide (SO2) and other trace gases contained in volcanic plumes. The standard spectral retrieval methods assume that all measured light comes from behind the plume and has passed through the plume along a straight line. However, a fraction of the light that reaches the instrument may have been scattered beneath the plume and thus has passed around it. Since this component does not contain the absorption signatures of gases in the plume, it effectively “dilutes” the measurements and causes underestimation of the gas abundance in the plume. This dilution effect is small for clean-air conditions and short distances between instrument and plume. However, plume measurements made at long distance and/or in conditions with significant atmospheric aerosol, haze, or clouds may be severely affected. Thus, light dilution is regarded as a major error source in DOAS measurements of volcanic degassing. Several attempts have been made to model the phenomena and the physical mechanisms are today relatively well understood. However, these models require knowledge of the local atmospheric aerosol composition and distribution, parameters that are almost always unknown. Thus, a practical algorithm to quantitatively correct for the dilution effect is still lacking. Here, we propose such an algorithm focused specifically on SO2 measurements. The method relies on the fact that light absorption becomes non-linear for high SO2 loads, and that strong and weak SO2 absorption bands are unequally affected by the diluting signal. These differences can be used to identify when dilution is occurring. Moreover, if we assume that the spectral radiance of the diluting light is identical to the spectrum of light measured away from the plume, a measured clean air spectrum can be used to represent the dilution component. A correction can then be implemented by iteratively subtracting fractions of this clean air spectrum from the measured spectrum until the respective absorption signals on strong and weak SO2 absorption bands are consistent with a single overhead SO2 abundance. In this manner, we can quantify the magnitude of light dilution in each individual measurement spectrum as well as obtaining a dilution-corrected value for the SO2 column density along the line of sight of the instrument. This paper first presents the theory behind the method, then discusses validation experiments using a radiative transfer model, as well as applications to field data obtained under different measurement conditions at three different locations; Fagradalsfjall located on the Reykjanaes peninsula in south Island, Manam located off the northeast coast of mainland Papua New Guinea and Holuhraun located in the inland of north east Island.