Previous published measurements of mercury photoreduction are for net-photoreduction, since photooxidation processes occur simultaneously. In this research we combine continuous dissolved-gaseous mercury (DGM) analysis with a photoreactor and a quartz sparger in order to derive mercury gross photoreduction rate constants for UVB and UVA irradiations. The DGM concentration in each filter-sterilized freshwater was measured at 5 min intervals over a period of 23 h. Photoreduction proceeded for the initial 200 min, after which, reducible mercury was depleted in the sample. Substantial losses in DOC fluorescence were observed during the incubations for UVA radiation but not for UVB; therefore, UVB photoreduction dynamics are not linked to a loss in DOC fluorescence. Pseudo first-order reaction kinetics fit the data well (r2 > 0.87). The rate constants appear divided between lakes and rivers with the mean lake UVB rate constant (kUVB = 8.91 x 10(-5) s(-1)), significantly less than the mean rate constant (kUVB = 1.81 x 10(-4) s(-1)) for the river samples. However, while there were differences for the UVB rates between lakes and rivers, the mean and median rate constants for UVA in lakes (kUVA = 7.76 x 10(-5) s(-1)) did not differ significantly from the mean rate constant forthe river sites (kUVA = 1.78 x 10(-4) s(-1)). Here, we propose a model for mercury photoredox dynamics for both temperate lake and river systems. The lake model was validated using principal axis analysis to compare observed and predicted DGM data (n=279) from a variety of lake sites in Nova Scotia and Central Quebec. Principal axis analysis found a linear fit (correlation = 0.81; slope = 2.13) between predicted and observed environmental DGM values when log-normalized. The constant bias on the predicted values was attributed to estimates of available reducible mercury and the effect of DGM volatilization on observed data.