Porosity is an important factor in the healing of prosthetic devices. To better understand this phenomenon, porous polyurethane scaffolds were produced by a variation of the phase inversion/porogen extraction technique in which a prepacked column of spherical porogen particles was infiltrated with a polymer solution before polymer precipitation and porogen extraction. Scaffolds contained pores of well defined shape (approaching open faced pentagonal dodecahedra), narrow size distributions (66.1 +/- 1.3 microm, 84.2 +/- 1.7 microm, and 156.9 +/- 1.2 microm) and high interconnectivity (interconnecting windows of 30.1 +/- 0.8 microm, 41.9 +/- 1.5 microm, and 76.4 +/- 2.0 microm, respectively). A high degree of accessible macroporosity (>80%) could be achieved while limiting the mostly inaccessible microporosity to below 2%. The neovascularization and inflammatory responses to the scaffolds were evaluated in the subcutaneous rat model for 4 weeks. The inflammatory response index and foreign body giant cell index could be reduced by 56% (p < 0.05) and 21% (p < 0.02), respectively, when the pore size was increased from 66 microm to 157 microm, whereas the vascularization index and arteriolar index remained unchanged. Thus, a significant decrease in inflammatory response could be achieved without adversely affecting the degree of neovascularization by increasing the size of the pores.