Data from 96 headwater lakes from Norway are used to model heavy metal concentrations in surficial lake sediments in relation to atmospheric deposition. The study evaluates the application of sediment-water partitioning models at the field scale and finds optimum values for the partition coefficients. The impact of environment (sediment type, lake water pH, etc.) on KDvalues is explored directly by comparing KDestimates with environmental variables. KDvalues for each metal are found by optimising the fit between predicted and observed surface-sediment concentrations. The sensitivity of the KDestimates to data structure is examined by bootstrapping. KDvalues of 105.8and 106.2were calculated for cadmium (Cd) and lead (Pb), respectively, comparable to recent direct observations. Biogenic silica influenced KDvalues for Cd, Pb and Zn, whereas lake depth influenced Pb. pH did not have any detectable effect. KDfor zinc (Zn) was less well defined, but higher than indicated by published experimental measurements. The results suggest that sediment-water partitioning models have an important contribution to make to field-scale lake studies of sediment heavy metals, and have important implications for palaeolimnological evaluations of heavy metal deposition.