Revegetation plantings are a key activity in farmland restoration and are commonly assumed to support biotic communities that, with time, replicate those of reference habitats. Restoration outcomes, however, can be highly variable and difficult to predict; hence there is value in quantifying restoration success to improve future efforts. We test the expectation that, over time, revegetation will restore bird communities to match those in reference habitats; and assess whether specific planting attributes enhance restoration success. We surveyed birds in 255 sites in south‐east Australia, arranged along a restoration gradient encompassing three habitat types: unrestored farmland (paddocks), revegetation plantings (comprising a chronosequence up to 52 years old) and reference habitats (remnant native vegetation). Surveys were undertaken in 2006/2007 and again in 2019, with data used to compare bird assemblages between habitat types. We also determined whether, in the intervening 12 years, bird communities in revegetation had shifted toward reference habitats on the restoration gradient. Our results showed that each habitat contained a unique bird community and that, over time, assemblages in revegetation diverged away from those in unrestored farmland and converged toward those in reference habitats. Two planting attributes influenced this transition: the bird assemblages of revegetation were more likely to have diverged away from those of unrestored farmland (with scattered mature trees) 12 years later if they were located in areas with more surrounding tree cover, and were mostly ungrazed by livestock (compared with grazed plantings). Our results highlight three key ways in which revegetation contributes to farmland restoration: (1) by supporting richer and more diverse bird assemblages than unrestored farmland, (2) by enhancing beta diversity in rural landscapes through the addition of a unique bird community, and (3) by shifting bird assemblages toward those found in reference habitats over time. However, revegetation plantings did not replicate reference habitats by ~40–50 years in our region, and complete convergence may take centuries. These findings have implications for environmental offset programs and mean that effective conservation in farmland environments depends on the retention and protection of natural and seminatural habitats as a parallel management strategy to complement restoration.