Terrestrial communities are strongly affected by habitat area and successional stage. As generalist predators, spiders represent an important community component that may be particularly sensitive to these landscape features. We used pitfall traps to sample ground-dwelling, cursorial spider abundance at 2 successional stages on an experimentally fragmented old-field site in northeastern Kansas. The site, established in 1984, includes replicate habitat patches undergoing secondary succession separated by a matrix of low turf maintained by regular mowing. Spiders were collected from traps located in large (5000 m²) and small (32 m²) patches during 1988 and 1990 (early succession), and 2005 (mid-succession). Total numbers collected were about 20% lower in 2005 than in 1988 and 1990. However, in all 3 yr significantly more spiders were collected by traps in large patches than by traps in small patches. The positive density–area relationship did not differ among years (no significant patch-size by year interaction), despite substantial change in patch vegetation during the 17 yr spanned by this study. The results for patch size are similar to those obtained in a study of insects sampled by sweep-net transects at the same site, which contrast with the negative density–area relationship reported for some small mammal species. The form of the density–area relationship may be determined by the size of the animals relative to the spatial scale of habitat fragmentation.