A critical need exists for quantitative reconstructions of long-term El Niño Southern Oscillation (ENSO) variability in the eastern tropical Pacific. Presented here is a method to quantitatively estimate past changes 1) in the seasonal amplitude of sea surface temperature (SST) in the Peruvian coastal upwelling system and 2) in the amplitude of ENSO-related interannual variability in the eastern tropical Pacific. The seasonal amplitude of SST (δT) along the length of the Peruvian coast is strongly correlated with the Niño1+2 index. We show that the frequency distribution of δT values provided by a modern sample of 13 Mesodesma donacium shells faithfully reflects modern ENSO variability at the regional scale, including the range of anomalies from La Niña to moderate El Niño events, but excludes extreme warm anomalies because of high shell mortality. We propose to use the frequency distribution of ENSO anomalies in paleoclimate studies for comparisons between shell records, coral records, and GCM simulations. Reconstruction uncertainties can be quantified using Monte Carlo simulations. The method presented here opens new perspectives for quantitative paleo-ENSO reconstructions in the Eastern Pacific since it may be applied with any mollusk species from Peru provided at least one annual cycle of SST is faithfully recorded by shell δ18O. © 2012 Elsevier B.V.