Gaps, 10 mm wide, in rat sciatic nerves were bridged by bioartificial nerve grafts consisting of a silicone tube containing seven longitudinally placed synthetic filaments, which were expected to serve as a scaffold for axonal growth. The filaments were made of non-resorbable material (polyamide [Ethilon (R)]) or resorbable material (polydioxanon [PDS (R)], polyglactin [Vicryl (R)] or catgut). The purpose was to study the influence of resorbable materials on axonal regeneration and to choose, in the long term, the best filament material among the four. After 3 and 6 months, histological techniques were used to study the regenerated nerve structure. The total axon number in the nerve segment distal to the silicone chamber was counted in all specimens at 6 months. The histological findings were different depending on the filament materials; all the three resorbable materials showing significantly larger numbers of axons than polyamide (non-resorbable). All materials were covered with several layers of more or less flattened cells. These results indicate that resorbable filaments are superior to non-resorbabIe filaments when used as a scaffold inside a silicone tube, and polyglactin seems ideal for this purpose. (C) 1997 Elsevier Science Ireland Ltd.