Stream restoration is often employed in efforts to stabilize eroding channel banks. Banks are stabilized through a designed channel approach, which involves grading and armoring of stream banks using heavy machinery, or alternatively through planting of seedlings and saplings to establish forested riparian buffers. We hypothesized that designed channel restoration would have detrimental impacts on riparian soils but that soils would recover over time, and we hypothesized that riparian buffer restoration would not impact riparian soils. We tested these hypotheses by comparing soil attributes (bulk density, soil organic matter, and root biomass) at reaches that had undergone designed channel and riparian buffer restoration in different years (project ages ranged from 2 to 16 years) to paired urban (unrestored) control reaches. Soil properties in sub-surface soil layers (10–20 and 20–30 cm depth) at both recent (<10 years old) and older (>10 years old) designed channel reaches differed significantly from paired urban control soils; bulk density was higher and root biomass lower in manipulated reaches compared to urban control reaches. At many designed channel reaches, bulk density exceeded values known to restrict root growth. These results indicate that compaction and disturbance of riparian soils may be a significant unintended consequence of designed channel restoration and can persist for at least a decade. In contrast, we found no significant differences in soil properties between riparian buffer restoration reaches and urban control reaches. Thus, the results indicate that riparian buffer restoration is a more ecologically favorable method than designed channel restoration for bank stabilization.