Environmental impact assessments require pre-impact data to discriminate between natural and anthropogenic sources of community shifts. True pre-impact data, however, are rarely available because the onset of ocean biomonitoring in the 20th century frequently postdates the timing of such shifts. Death assemblages (DAs) sampled on a sea-floor capture previous community states and represent archives of past preimpact conditions. This is because they change more slowly (10–10,000 years) than the corresponding living assemblage (LA) (yearly scales of turnover). Strong and rapid directional changes such as those due to human activities are therefore not immediately captured by DAs, leading to a greater live-dead (LD) mismatch than under natural processes alone. We used this approach to infer disturbance of benthic habitats around platforms in two major oilfields in the Persian (Arabian) Gulf: Umm Al Dalkh (UA) and Zakum (ZK). We found that, first, LD metrics showed no correlation with contamination levels or dista