Long-term chronologies of precipitation can provide a baseline against which twentieth-century
trends in rainfall can be evaluated in terms of natural variability and anthropogenic influence. However,
there are relatively few methods to quantitatively reconstruct palaeoprecipitation and river discharge
compared with proxies of other climatic factors, such as temperature. We developed autoregressive and
least squares statistical models relating Chesapeake Bay salinity to river discharge and regional
precipitation records. Salinity in northern and central parts of the modern Chesapeake Bay is
influenced largely by seasonal, interannual and decadal variations in Susquehanna River discharge,
which in turn are controlled by regional precipitation patterns. A power regressive discharge model and
linear precipitation model exhibit well-defined decadal variations in peak discharge and precipitation. The
utility of the models was tested by estimating Holocene palaeoprecipitation and Susquehanna River
palaeodischarge, as indicated by isotopically derived palaeosalinity reconstructions from Chesapeake Bay
sediment cores. Model results indicate that the early/mid Holocene (7055/5900 yr BP) was drier than the
late Holocene (1500 yr BP / present), the ‘Mediaeval Warm Period’ (MWP) (1200/600 yr BP) was drier
than the ‘Little Ice Age’ (LIA) (500/100 yr BP), and the twentieth century experienced extremes in
precipitation possibly associated with changes in ocean/atmosphere teleconnections.