This study utilized freshwater mussel shells to assess mercury (Hg) contamination in the North Fork Holston River that extirpated (caused local extinctions of) a diverse mussel fauna. Shells (n = 366) were collected from five sites situated upstream (two sites), just below (one site), and downstream (two sites) of the town of Saltville, Virginia, where Hg was used to produce chlorine and caustic soda from 1950 to 1972. Shell samples were used to test the (1) utility of geochemical signatures of shells for assessing the spatial variation in Hg levels in the river relative to the contamination source and (2) value of taphonomy (post-mortem shell alteration) for distinguishing sites that differ in extirpation histories. Geochemical signatures of 40 shells, analyzed using atomic absorption spectroscopy, indicated a strong longitudinal pattern. All shells from the two upstream sites had low Hg concentrations (<5-31 microg/kg), shells directly below Saltville had variable, but dramatically higher concentrations (23-4637 microg/kg), and shells from the two downstream sites displayed intermediate Hg levels (<5-115 microg/kg) that declined with distance from Saltville. Two pre-industrial shells, collected at Saltville in 1917, yielded very low Hg estimates (5-6 microg/kg). Hg signatures were consistent among mussel species, suggesting that Hg concentrations were invariant to species type; most likely, highly variable Hg levels, both across sites and through time, overwhelmed any interspecific differences in Hg acquisition. Also, a notable postmortem incorporation of Hg in mussel shells seemed unlikely, as the Hg content was not correlated with shell taphonomy (r= 0.18; p = 0.28). The taphonomic analysis (n = 366) showed that the degree of shell alteration reliably distinguished sites with different extirpation histories. At Saltville, where live mussels have been absent for at least 30 years, shells were most heavily altered and fragmented. Conversely, fresh-looking shells abounded upstream, where reproducing mussel populations are still present. In summary, relic shells offered valuable spatio-temporal data on Hg concentrations in a polluted ecosystem, and shell taphonomic signatures discriminated sites with different extirpation histories. The shell-based strategies exemplified here do not require sampling live specimens and may augment more standard strategies applied to environmental monitoring. The approach should prove especially useful in areas with unknown extirpation and pollution histories.