<p><strong>Abstract.</strong> Future deep-sea mining for polymetallic nodules in abyssal plains will impact the benthic ecosystem, but it is largely unclear whether this ecosystem will be able to recover from mining disturbance and if so, at what time scale and to which extent. In 1989, during the &quot;DISturbance and reCOLonization&quot; (DISCOL) experiment, a total of 22&#8201;% of the surface within a 10.8&#8201;km<sup>2</sup> large circular area of the nodule-rich seafloor in the Peru Basin (SE Pacific) was ploughed to bury nodules and mix the surface sediment. This area was revisited 0.1, 0.5, 3, 7, and 26 years after the disturbance to assess macrofauna, megafauna and fish density and diversity. We used this unique abyssal faunal time series to develop carbon-based food web models for disturbed (sediment inside the plough tracks) and undisturbed (sediment inside the experimental area, but outside the plough tracks) sites. We developed a linear inverse model (LIM) to resolve carbon flows between 7 different feeding types within macrofauna, megafauna and fish. The total faunal biomass was always higher at the undisturbed sites compared to the disturbed sites and 26 years post-disturbance the biomass at the disturbed sites was only 54&#8201;% of the biomass at undisturbed sites. Fish and sub-surface deposit feeders experienced a particularly large temporal variability in biomass and model-reconstructed respiration rates making it difficult to determine disturbance impacts. Deposit feeders were least affected by the disturbance, with respiration, external predation and excretion levels only reduced by 2.6&#8201;% in the sediments disturbed 26-years ago compared with undisturbed areas. In contrast, the respiration rate of filter and suspension feeders was still 79.5&#8201;% lower after 26 years when comparing the same sites. The &quot;total system throughput&quot; (<i>T</i>..), i.e. the total sum of carbon flows in the food web, was always higher at undisturbed sites compared to the corresponding disturbed sites and was lowest at disturbed sites directly after the disturbance (8.63&#8201;&#215;&#8201;10<sup>&#8722;3</sup>&#8201;&#177;&#8201;1.58&#8201;&#215;&#8201;10<sup>&#8722;5</sup>&#8201;mmol&#8201;C&#8201;m<sup>&#8722;2</sup>&#8201;d<sup>&#8722;1</sup>). Therefore, 26 years after the DISCOL disturbance, the throughput discrepancy between the undisturbed and the disturbed sediment was still 56&#8201;%. From these results we conclude that C cycling within the fauna compartments of an abyssal plain ecosystem remains reduced 26 years after physical disturbance, and that a longer period of time is required for the system to recover from such a simulated small scale deep-sea mining experimental disturbance.</p>