Urbanization is associated with substantial losses to stream biological diversity throughout the United States' mid-Atlantic. Stream restoration has been used to improve stream conditions and, in part, to ameliorate these losses. However, the relationship between restoration and recovery of biological diversity is unclear. Our objective was to critically examine the efficacy of urban stream restorations with regard to biological diversity. We compared restored urban streams to urban nonrestored, nonurban, and reference (minimally degraded) streams using five measures each of fish and benthic macroinvertebrate diversity. Both multivariate and univariate statistical analyses show biological diversity of restored urban streams to be similar to nonrestored urban streams and lower than nonurban and reference streams. Restored urban sites showed no apparent increase in biological diversity through time, while diversity decreased at two of the reference streams coincident with increased urban development within their catchments. Our results indicate that restoration approaches commonly used regionally as in these urban streams are not leading to recovery of native stream biodiversity. Evidence from several sources indicates a need for dramatic changes in restoration approach, and we argue for a watershed-scale focus including protection of the least impacted streams and adopting other land-based actions within the watershed where possible.