Task of the present Master thesis is the analysis of different geophysical methods, like magnetics, susceptibility, electromagnetics, self-potential and geoelectric measurements, in respect of their responsiveness on near surface gas leakages. The advantage of these geophysical methods is, that they are non invasive or rather less invasive than the state-of-the-art technology in exploration of gas leakage today. In the process of locating a gas leak, it is always necessary to penetrate the ground and to vacuum the gas to the measuring system. So we can never talk about an undisturbed ground, in connection with gas leak detection. The main objective of this work is to evaluate new ways for the detection of near surface gas leaks, or even to get a better understanding of the behaviour of the processes that happen when a gas leak occurs. The measurements were performed at the testing facility of the EVN AG in Tulln, to guarantee the reproducibility of the experiment. There it was possible to perform all the geophysical measurements under controlled conditions, like a known amount of gas or perception. The methodology contains a comparison of different geophysical methods. The experimental part of this work comprises the collection and analysis of measurement data from the geophysical methods in addition with information gathered by the measurement of the gas concentration at the surface. The results of the field-measurements showed that only the self-potential method can be used for the direct detection of near surface gas leaks, and for some applications this method could be a good and cheaper alternative to normal gas detection systems. Another method which led to new insights was the multielectrode-geoelectrics. The information gathered by these measurements led to a completely new viewing about the influence of perception and the time to pause the measurements, needed after a rainfall. The outcome of the field-measurements together with previous studies can help to avoid inaccurate measurements and to make gas leak detection more efficient.