Many factors are suspected to bias the composition of death or fossil assemblages relative to their source living communities, thus complicating their use in paleoecology. Combining abundance information from 17 live-dead datasets for subtidal bivalves (various authors) with published data on species lifespans, we evaluate bias of proportional abundance information owing to differences in mortality rate. Mortality bias (the difference between the expected dead abundance and the living abundance) does change significantly from positive to negative with increasing lifespan, but its effect is remarkably small (mean change = ~2 percentage points, median <1%). Proportional abundance in the expected death assemblage is positively correlated with live abundance, with a slope not significantly different from unity: a death assemblage biased only by differential mortality would not differ detectably from a snapshot of the living standing crop. Thus in broad-scale analyses of bivalve faunas, differences among species in lifespan and mortality rate add only (slight) noise rather than bias. Observed bias (the difference between a species' proportional abundance in the death assemblage and its proportional abundance in the living assemblage) does not vary with lifespan, indicating that other sources of bias (e.g., post-mortem transport, differential destruction of species) compensate for the slight positive mortality bias seen among short-lived species. However, observed bias is also low (average ~6%, median <1%), is focused on the most abundant species, and tends to reduce their proportional abundance. Overall, death assemblages thus tend to have higher evenness, but preserve information on richness, species composition, and species rank order.