The karst lake district of peninsular Florida shares hydrologic properties with certain glaciated districts, such as outwash plains, but differs from most well‐watered regions in losing much more water to subsurface effluents than to surficial runoff. In nutrient‐loading models that begin as hydrologic throughflow models, groundwater inflow and outflow are commonly ignored or estimated from errors of closure of water budgets. Such models are inapplicable to seepage lakes, unless deep‐seepage outflow can be measured. Estimates of downward leakage from Florida lakes with and without outlets suggest that conventional nutrient‐loading models are inapplicable to any lake not in a watertigh rock basin.Downward leakage is measurable during months when lake level falls by an amount exceeding the net precipitation deficiency. To measure it, USGS stage data for 20 lakes from north, central, and south sectors of the peninsula were compared with three 32‐yr sets of mean monthly meteorologic measurements. Monthly pan‐evaporation data were converted to lake evaporation by factors estimated at Lake Okeechobee. Leakage estimates (geometric means) range from 28.4 to 50.9 cm yr‒1 in 18 of 20 lakes; two exceptionally astatic lakes gave 90.1 and 141.2 cm yr‒1 . Presence of outlets in 15 of 20 lakes makes no detectable difference in leakage. Adding evaporative losses (and ignoring surface outflow if any), residence times for the 20 lakes averaged 2.67 ± 1.33 yr.Florida lake levels fluctuate in sympathy with each other and respond with little or no lag to monthly net precipitation. Longer term swings are also evident and are more highly correlated with artesian pressure in the deep limestone aquifer than with meteorologic variables. Few if any lakes receive direct injections from this aquifer, and artesian influence is exerted on lakes and local water tables through many meters of surficial (Mio‐Pliocene) deposits.