A linked 1-dimensional thermal-dissolved oxygen model was developed and applied in the central basin of Lake Erie. The model was used to quantify the relative contribution of meteorological forcings versus the decomposition of hypolimnetic organic carbon on dissolved oxygen. The model computes daily vertical profiles of temperature, mixing, and dissolved oxygen for the period 1987–2005. Model calibration resulted in good agreement with observations of the thermal structure and oxygen concentrations throughout the period of study. The only calibration parameter, water column oxygen demand (WCOD), varied significantly across years. No significant relationships were found between these rates and the thermal properties; however, there was a significant correlation with soluble reactive phosphorus loading. These results indicate that climate variability alone, expressed as changes in thermal structure, does not account for the interannual variation in hypoxia. Rather, variation in the production of organic matter is a dominant driver, and this appears to have been responsive to changes in phosphorus loads.