Summary The establishment of direct seeded revegetation is well researched. However, there is little understanding of whether revegetation simplifies with age and loses many of the short-lived understorey shrub species that provide critical resources for birds and other fauna, or regenerates sufficiently to be self-sustaining. We sought to address this by investigating the change in structure and composition of 33 direct seeded sites established by Greening Australia between 1990 and 1996 in the Southern Tablelands of NSW. Transects were used to collect data describing the abundance and richness of woody plants in 1998 and 2008, and the abundance of woody plant regeneration in 2008. Our analysis showed the predicted number of live stems per metre declined exponentially from 5.8 stems/m (∼17 000 stems/ha) at age 1.5 years to 1.5 stems/m (∼4500 stems/ha) at age 17.5 years. Predicted woody plant species richness also declined with age, with a linear relationship. The number of species in the seed mix affected predicted woody plant species richness. However, large increases in seed mix richness produced relatively small increases in predicted species richness. Regeneration (new stems) was present at high levels (>100 stems/ha); however, 82% of regeneration appeared to originate from delayed germination of sown seed rather than seed from established plants (recruitment). The predicted abundance of new stems (delayed germination and recruitment) declined with age, and for a given age increased with row width. Young stands (11.5 years), seeded with wide rows (4 m), had approximately six times the new stems of similar aged stands seeded with narrow rows (2 m). Our results indicate direct seeded stands simplify with age, becoming less dense and containing fewer species. Maintaining a diversity of shrub species in direct seeded sites may therefore require ongoing disturbances (scarifying, scalping, fire, thinning) of established sites and changes to establishment techniques for new sites. We suggest further research to establish the status of soil seed banks in direct seeded sites, testing different forms of disturbances to trigger regeneration in older revegetation sites and establishing new sites using wide rather than narrow rows and seed mixes enriched with species from genera other than Eucalyptus or Acacia.