It is well known that in aquatic environment inorganic and organic contaminants behaviour is in a great extent controlled by the dissolved natural organic matter (DNOM), modifying their bioavailability and so their toxicity toward micro- and macro-organisms. The properties and reactivity of natural heterogenic DNOM could be strongly influenced by numerous physicochemical processes occurring in natural waters. A coastal zones like estuaries are of particular interest because of their susceptibility to the anthropogenic influence and a large spatio-temporal variations of physico-chemical conditions (pH, I, T, Eh, c). The prediction of the temporal behaviour of contaminants as well as their distribution in such exchanged physico-chemical conditions is of particular interest. Thus a combination of kinetic and equilibrium approach in DNOM-pollutant interactions studies should be preferential. The characterisation of the DNOM-pollutants interactions (by mathematical parameters usable in biogeochemical models) is very demanding due to very low content of both, contaminants (~ nmol.L-1) and organic matter (mg.L-1). A development of an appropriate analytical and modelling tools, as well as approaches, is a pathway that should be followed. Present study concerns to the investigation of DNOM-Cu2+ interactions in a non-polluted environment (Krka River estuary, ibenik, Croatia) by coupling both kinetic and equilibrium approaches. A vertical profile of water samples with different salinity was submitted for this study. The methodology was based on the measurement of non-organic Cu2+ fraction by applying differential pulse anodic stripping voltammetry (DPASV) and a mercury drop electrode. A copper titration experiments in a logarithmic mode (from nmol.L-1 to µmol.L-1) were performed. It is found that DNOM-Cu association, varying from freshwater to marine one, appears to be quite slow (~ 1- 2 h) probably due to competitive effect of major divalent cations (Ca2+, Mg2+). The obtained data were modelled on both kinetic and equilibrium model. A home-written non-linear fitting program allowed calculation of two sets of final data (a ligand concentration and a conditional stability constant), corresponding to these two models. A very good agreement between these two sets of results were obtained.