One major issue in restoration ecology is the extent to which restorations mimic the composition and function of remnant areas. We monitored plant species diversity in 4 restored sites at The Nature Conservancy’s Efroymson Prairie Restoration in Newton Co., Indiana, and in 3 remnant prairie communities surrounding the restoration. Data on plant richness and abundance were collected for 4 years for 3 restoration sites and 2 years for 1 restoration and the remnant prairies.  Compared to remnant prairies, the restored sites had a lower proportion of native species and lower species diversity; mean Shannon diversity (1 stdev) for restored sites = 3.6 (0.4) vs. remnant sites = 4.3 (0.3).  Restored sites had greater variability in the ratio of forbs to grasses (restored 15-75% forbs vs. remnant 58-65% forbs). Species richness increased 36-50% in two restored sites over four years and by 16% in one site that was monitored for 2 years. Species richness declined by 10% over 4 years in the oldest restored site (planted in 1997).  The Chao-Jaccard (CJ) abundance-based estimator compares species abundance between sites and reduces bias due to undersampling or uneven species distribution. Comparison of the abundance-based Jaccard index with the CJ estimator suggests that compositional similarity estimates were more biased in restored sites compared to remnant sites, perhaps due to greater numbers of rare species. Our data also show that species abundances chosen for the initial planting persisted for at least 8 years.