A fracture simulation model which incorporates the physics of fracture growth is used to investigate how the mechanics of fracture formation affect the flow characteristics of fractured systems. Fractures are assumed to grow subcritically with the growth rate given by a power law function of the energy available for fracture growth. Flow characteristics are quantified in terms of the percent of networks percolating and the average effective conductivity as a function of the fracture density. For all flaw densities considered and for values of the growth rate exponent &agr;≤1, the flow characteristics primarily depend on the fracture spatial density and are similar to the flow characteristics of networks generated stochastically by assuming the fractures are randomly located. For &agr;≤1, the mechanical interaction of the flaws and fractures imparts an organized structure to the network resulting in isolated fractures, or zones of fractures, which form extensive, connected pathways at significantly lower fracture densities. Experimentally measured values of &agr; for subcritical fracture growth are typically greater than one, suggesting that the flow characteristics of randomly located fractures may not be representative of natural fracture networks thought to have grown subcritically. An analysis of published fracture trace maps suggests that many natural fracture networks have fracture spatial densities near the percolation threshold. It is suggested that this may be due to the existence of a self-limiting mechanism in fracture network formation. ¿ American Geophysical Union 1996