The eutrophication problem has drawn attention to nutrient leaching from arable land in southern Sweden, and further understanding of spatial and temporal variability is needed in order to develop decision-making tools. Thus, the influence of spatial and temporal variables was analysed statistically using empirical time series of different nutrient species from 35 well-documented catchments (2-35 km(2)), which have been monitored for an average of 5 years. In the spatial analysis several significant correlations between winter median concentrations and catchment characteristics were found. The strongest correlation was found between inorganic nitrogen and land use, while concentrations of different phosphorus species were highly correlated to soil texture. Multiple linear regression models gave satisfactory results for prediction of median winter concentrations in unmeasured catchments, especially for inorganic nitrogen and phosphate. In the analysis of temporal variability within catchments, internal variables from a dynamic hydrological model (HBV) were linked to concentration fluxes. It was found that phosphorus and inorganic nitrogen concentrations were elevated during flow increase at low-Bow conditions, while they were diluted as the wetness in the catchment increased. During unmonitored periods regression models were successful in predicting temporal variability of total phosphorus, phosphate and inorganic nitrogen, while organic nitrogen and particulate phosphorus could not be predicted with this approach. Dividing the data into different flow categories did not improve the prediction of nutrient concentration dynamics. The results and literature review presented, confirm parts of the present HBV-W model approach and will be useful for further development of nutrient routines linked to dynamic hydrological models. (C) 2000 Elsevier Science B.V. All rights reserved.