Representative sampling is of great importance to decision making regarding contaminant site risks and remedial effectiveness. A focus here is whether existing groundwater monitoring wells screened across the water table can be sampled to yield representative indicators of soil gas composition. For the first time, we provide multi-phase, multi-component computational simulations to address this. We simulated high and low gas extractions rate strategies to sample the gas phase over short and extended screening intervals across the water table. We investigated the options against a field data set representative of typical hydrocarbon vapour profiles, inclusive of major gases, oxygen and carbon dioxide. We also evaluated the sampling options for uniform and non-uniform multi-component gasoline LNAPL distributions, including hazardous chemicals. Less sensitivity to the sampling option was observed for depth-wise increasing concentration profiles with a near-constant concentration across the screen. Significant discrepancy between the ratio of different compounds in the sample and in-situ (real) values was observed for high-rate gas extraction (particularly for an extended-screen). Low-rate gas extraction provided satisfactory results for all the scenarios. Shorter screening slightly improved the accuracy of this option. Graphical representations are provided to allow assessment of the applicability of each sampling option for various site conditions. Crown Copyright (C) 2017 Published by Elsevier B.V. All rights reserved.