The proposed paper offers an alternative conceptual model for the migration of 90Sr from a shallow radioactive waste trench to the aquifer based on a detailed interpretation of geochemical field observations and subsequent testing of the hypothesis through modelling. The results indicate that a rising water table and subsequent flooding of the trench bottom contributes to the release of 90Sr to the aquifer and may help to explain the transient concentration changes observed. This contribution is added to the permanent flux of 90Sr transport through the unsaturated zone to the aquifer as the consequence of infiltrating rain water. The study concerns a shallow waste trench located at 2.5 km west of the Chernobyl Nuclear Power Plant (NPP). It contains mostly organic debris coming from irradiated trees and fuel particles deposited on the surface after the explosion. It has a very heterogeneous content, is about 80 m long and 10 m wide and has a variable depth. The first global modelling exercise on the migration of 90Sr at the experimental Chernobyl Pilot Site (CPS;(Bugai et al., 2005)) indicated some discrepancy between model predictions and field observations on the form of the 90Sr plume. The model assumed infiltration of 90Sr with infiltrating rainwater through the trench and the unsaturated zone in order to reach the aquifer. However, interpretation of data gathered at a high water table level, a subsequent lower water level and a third higher water level with Principal Component Analysis (PCA) and other tools indicates that water table fluctuation plays a role in the 90Sr amount released. This hypothesis suggests the following (1) a rise in water level submerging the bottom of the trench releases radionuclides, at the same time a number of other elements typical for the trench but not the surrounding aquifer also dissolve in the water (e.g. Ca2+, SO42-, K+ , NO3 - etc) (2) when the water table lowers below the trench bottom, the concentration of radionuclides and other aquifer foreign elements are diluted by the simple action of less charged upstream water flux and infiltrating rainwater. The main conclusion indicates that as noted by Bugai et al. (2005) the distribution coefficient (Kd) is difficult to fix as chemical conditions change over time. The correspondence with laboratory derived Kd values in batch or column experiments at constant chemistry is therefore not expected. To confirm this hypothesis, geochemical modelling using the 90Sr isotherm and derived exchange capacity of the sediments present at site is carried out.