Evaluation of tidal marsh restoration success is typically based on the recovery of habitat size and target species. However, food-web structure may provide valuable insight into ecosystem functioning trajectories. Here, we studied restored tidal marshes of different ages (new, young, old; spanning 1–150 years) in comparison with nearby reference sites along the San Francisco Estuary. We asked: (1) How does restoration help recover energy pathways that support fishes? (2) Do fishes rely more on algal versus detrital pathways in restored sites?; and (3) How does food-web structure vary as a function of species origin and life history? To answer these questions, we sampled fish (n = 806) and basal resources (emergent vegetation and phytoplankton; n = 109) seasonally over two hydrologically contrasting years. Using stable isotopes (δ¹³C, δ¹⁵N, and δ³⁴S), we calculated fish isotopic niche volumes, food chain lengths, and the relative importance of algal versus detrital energy pathways. We found that food chains in restored sites were 8% shorter than in their paired reference sites. Additionally, the young and old restored sites had 37% smaller niche volumes than their references, but the opposite was true for the new restored site (11% larger), illustrating the characteristic trophic surge of early succession. Fishes found in restored sites relied significantly less on detrital energy (7% less) than fishes found in reference sites, and resident fishes showed 12% higher reliance on the detrital pathway than transient species. Finally, most of the native niche volume overlapped with that of introduced fish, which was in turn 38% larger, and a similar pattern was observed when comparing resident to transient fish. Our findings demonstrate that food-web structure does not immediately recover with tidal marsh restoration, even if fish assemblages are species-rich; and show that transient trophic surges may complicate restoration success assessments of newly restored marshes. We contend that incorporating recovery of energy pathways as an indicator of performance may help strengthen monitoring and design of wetland ecosystem restoration projects.