The general aspects of the Edwards Aquifer have been understood for over a century - losing streams are the most important source of recharge, discharge is concentrated in a few major springs, and the Edwards Limestone (and equivalent formations) host the aquifer. However, as new data emerged, numerical models of the aquifer and its recharge have evolved in the past 4 decades and raised new questions. Initial models of the San Antonio and Barton Springs segments of the aquifer relied on a limited stream gauges to assess recharge. Direct recharge on aquifer outcrops was calculated as the difference between recharge and gauged spring discharge. Direct measurements of evapotranspiration and precipitation in recharge areas showed that direct recharge is more significant than first modeled. Detailed stream gauging showed that stream losses are concentrated in limited sites in the streambeds and stream loss calculation is more complex than assumed in most models. Recent analyses found significant additional discharge at Cold Springs in Austin and at Los Moras Springs and the Leona River alluvium in the San Antonio segment. Groundwater divides are observed to fluctuate spatially and some disappear during droughts. Tracer tests have redefined recharge areas. In some places, faulting has juxtaposed the underlying Glen Rose and the overlying Buda Limestone and Austin Chalk so that these formations are essentially part of the aquifer. Additional recharge has been shown to come from the Trinity Aquifer, small ungauged creeks that flow over the recharge area, and the effects of urbanization. Models are one of our most important tools for assessing and managing groundwater systems, but experience in the Edwards has shown they require continual readjustment as new data become known. We should expect more alterations to our models (both conceptual and numerical) of the Edwards and other karstic aquifer systems and the incorporation of these into management models.