Although a number of metal speciation methods are nowadays available, most water quality regulations are based on total metal concentrations. One of the main reasons for the ignorance of speciation is the lack of methods with potentiality for monitoring. Conditions that have to be met by such speciation methods are: simple performance, robust and sufficiently accurate. In this study the potential for monitoring purposes of Diffusive Gradients in Thin films (DGT) is investigated as part of a European project for sensor development for routine prediction of metal biouptake in natural waters. Performance characteristics of DGT have been assessed in experiments under controlled laboratory conditions and during in situ application over 4-9 months under different hydrological conditions in the rivers Meuse and Rhine. Results have been worked out for Cu, Ni and Pb. The study shows that DGT has sufficient potential as a robust tool in routine monitoring to observe trends in water quality. However, it appears that during in situ application of DGT the measurement uncertainty of the results is much larger than under laboratory conditions. The increase in measurement uncertainties will partially be due to uncertainties in some as constant considered factors in the calculation of the DGT concentration. For Cu and Ni, the average values for the reproducibility in the rivers Meuse and Rhine appeared to be 28% and 17%, respectively, whereas under laboratory conditions the reproducibilities for both metals were better than 10%. In the speciation of Cu, Ni and Pb in the rivers Meuse and Rhine, the labile fraction, determined using DGT, decreased in the series Ni, Cu, Pb. The ratio of the non-labile and labile fractions of the metals appeared to decrease with increasing content of Dissolved Organic Carbon (DOC).