Numerical models of ground-water flow have met with limited success when applied to aquifers in soluble rocks, commonly referred to as karstic aquifers. Although modeling of karstic processes is often possible and numerical flow models can sometimes simulate hydraulic heads, ground-water fluxes, and spring discharge, they often fail to correctly predict such fundamental information as flow direction, destination, and velocity. The poor performance of numerical models in karstic aquifers has resulted from: 1) inappropriate emphasis on the local scale when flow paths are often orders of magnitude longer, 2) the large amount of hydrogeologic data required to model such anisotropic, heterogeneous aquifers and the additional data requirements necessary to establish realistic boundary conditions 3) the necessity to develop models which incorporate triple continuum concepts, 4) the difficulty of properly characterizing and modeling transient events in karst aquifers and, 5) the inadequacy of laminar flow equations or Darcy's law to describe the flow in conduits in the aquifer. Most numerical models of ground-water flow in karst aquifers nevertheless continue to employ single-continuum porous-medium models such as MODFLOW and are based on invalid interpolation of inappropriate data collected far too infrequently from points that did not intersect preferential flow paths. Unfortunately, some of these modeling attempts have been substituted for the collection of the additional data that are essential to providing adequate characterization of ground-water flow at sites underlain by karstic aquifers.