Summary Fencing remnant native vegetation has become a widespread activity for arresting declines in biodiversity in agricultural landscapes. However, few data are available on the effectiveness of this approach. The present study investigated the short-term effects of fencing to exclude livestock on dominant tree and shrub recruitment, plant species cover, litter and soil characteristics in remnant grassy woodlands in southern NSW. Vegetation and soil surveys were undertaken at 47 sites fenced by Greening Australia (NSW) for 2'4 years. Fenced and unfenced areas at each site were compared using split-plot sampling. Woodlands sampled were dominated by Yellow Box/Blakely's Red Gum (Eucalyptus melliodora/Eucalyptus blakelyi), Grey Box (Eucalyptus microcarpa) or White Cypress-pine (Callitris glaucophylla). Significantly higher numbers of tree recruits were found in the fenced sites, with tree recruitment found in 59% of fenced sites compared with 13% of unfenced sites. Fenced sites also had significantly greater cover of native perennial grasses, less cover of exotic annual species and less soil surface compaction. However, outcomes varied among woodland ecosystems and individual sites. Where tree recruitment occurred, there was significantly more tree recruitment where there was greater perennial grass cover and less regeneration where exotic annual grass cover or overstorey crown cover was dense. Few shrubs recruited in fenced or unfenced areas, reflecting the lack of mature shrubs in most sites. Fencing is an important first step for conserving threatened grassy woodlands, but more active management may be needed to enhance woodland recovery, particularly in sites where few or no recruits were found.