Based on presence/absence of species, we proposed previously that large fragments of the Florida sand pine scrub habitat do not possess greater conservation value for vertebrates than do archipelagos of small and medium-sized fragments. We reexamined the values of fragments of different sizes, based this time on abundances of species. We asked whether abundances tend to decrease with decreasing scrub size or are more strongly related to other scrub attribute(s). We censused vertebrates with trap arrays and by direct observation, and we measured 13 potentially important environmental attributes related to area, isolation, and habitat structure in 16 scrubs distributed along the Lake Wales Ridge of central Florida. We cor- related abundance of species with the 13 attributes. We calculated the evenness of relative abundance distributions and grouped scrubs based on the shapes of relative abundance distributions. We determined whether or not members of different groups of scrubs pos- sessed different values for any attribute. Finally, we examined both densities of individuals and dispersions of species among large, medium, and small scrubs. Abundance was related to many of the attributes. Abundances of 11 of 18 species were correlated positively with area. Percentage of variation explained by area generally was low, and several of the rare species apparently maintain relatively large populations in small scrubs. Abundances of five species were correlated negatively with measures of isolation. Abundances of three species were correlated both positively and negatively with measures of habitat structure. Environmental attributes related to isolation and habitat structure were correlated most strongly with both evenness and groupings of relative abun- dance distributions. Total density of individuals (i.e., numbers of captures per trap array) could not be shown to differ among scrubs of different sizes. Individually, the species that reached their highest densities in relatively small scrubs often were the ones considered to be rare. Spatial distributions of species tended to become more homogeneous as area decreased, but the strength of this effect varied among species, perhaps accounting, at least in part, for the different relationships between abundance and area among species. Our results could lead to broad generalizations that seem to be relevant to scrub con- servation. For example, because most species are relatively abundant in large scrubs, large scrubs should be selected for reserves over small ones. We suggest, however, that our results actually best serve as a caution about the use of generalizations, based on data from one or a few species, for conservation planning in general. Our results clearly show that the benefits gained for certain species from focusing on the preservation of larger scrub frag- ments could be offset by harm done to other species, especially rare species. We conclude, therefore, that selection of a system of scrub reserves should be based, in large part, on knowledge of the biologies of as many of the resident organisms as possible. We suggest, however, that any reserve selection procedure for the scrub habitat probably is outmoded, and that ecologists should not even engage in debate about the conservation value of small scrub fragments. When a habitat has declined as precipitously as scrub has, and truly large remaining fragments are, at best, extremely rare, then smaller fragments are likely to be of considerable value, no matter how they compare to larger fragments.