ABSTRACT Ten years of monthly water quality data were compared from two large shallow lakes in Florida, USA – Lakes Okeechobee and Apopka. Seasonal changes in trophic state index (TSI) values and log-log regression models relating total phosphorus (TP), total nitrogen (TN), chlorophyll a (CHLA), and Secchi transparency (SD) were considered. The objective was to quantify the extent to which empirical models might vary due to the heterogeneous nature of benthic-pelagic coupling that can occur in shallow lakes. In the offshore region of Lake Okeechobee, TP and SD-based TSI values increase dramatically during winter, while CHLA-based TSI declines. These changes coincide with the windy season in south Florida, when average wind velocities can exceed 20 km h−1. Resuspension of bottom sediments occurs and this reduces light penetration to the extent that the growth of phytoplankton is inhibited. During summer, winds are calm, and these conditions occur less often. Log-log regression models of CHLA vs. TP have a negative slope during winter and a positive slope during summer in offshore Lake Okeechobee. The slope of SD vs. CHLA models displays the opposite seasonal pattern. Seasonal changes are considerably muted in the near-shore region, where sediments are more consolidated and where shallower depths reduce the severity of light suppression of phytoplankton growth. The magnitude of seasonal variation in TSI values in Lake Apopka is muted, perhaps because seasonal variation in wind velocity is less than in the region of Lake Okeechobee. Hypereutrophic levels of nutrients, CHLA, and SD occur year-round. The log-log regression models of CHLA vs. TP and CHLA vs. TN are highly significant, and light availability does not appear to limit rates of algal production. Frequent resuspension events may occur, but negative effects on phytoplankton may be mitigated by: (1) shallower depth; (2) a greater presence of viable algae in the surface sediments that might “seed” the water column with phytoplankton; (3) high concentrations of soluble nutrients in sediments that might stimulate algal growth; and (4) lower concentration of light-attenuating inorganic particles than in Lake Okeechobee. The results indicate that a high degree of variability in both the seasonal patterns and relationships among nutrients, CHLA, and transparency can occur in shallow subtropical lakes, even when they share common features (e.g., geographic location, large size, wind patterns). This finding has practical implications regarding the use of regression models in the context of shallow lake management. Models may be quite useful on a lake-by-lake basis (e.g., in Lake Apopka they have successfully been used in eutrophication management), but considerable care should be taken when generalizing models to other lake ecosystems.