A suite of geochemical oil samples from a wide geographic area and a variety of depositional environments was analysed by INAA and ICP-MS. Good correlation between the two techniques was obtained for V, Ni and Co (r= 0.95–0.90). The correlation decreased in the order: Zn (r= 0.87), Fe, Sb, Se, U, Ti, La, As, Ba, Mn and W (r= 0.18). The contribution of trace elements to oils from produced waters, drilling fluids, biodegradation and migration has been suggested by some of the samples for which there was disagreement. Laboratory experiments confirmed that barite contributes Mn, Fe, Ga and Pb to oil samples that have not been filtered. On the ICP-MS 130Ba2+ interferes with 65Cu+ whereas 63Cu+ suffers from interference by 51V12C+. If formation water and produced water is not removed from the oil, the water contribution of elements e.g., As and Br is superimposed onto the organically bound contents in the oils. In such instances, it appears that these elements are not useful in fingerprinting and classifying oils. However, when the formation water contribution is removed, elements e.g., As are shown to be geochemically significant. It has thus been demonstrated that it is important to water wash and filter oil samples before analysing for geochemical exploration purposes. This work has also shown that metals such as Zn, Cd and Pb are picked up by oils migrating through ore bodies of these metals. Prior to examining the contributions of water, barite and migration, oils could only be classified using their V, Ni and Co contents. The new understanding acquired from this work, has made it possible to identify and eliminate samples with severe contamination. Application of multivariate statistics to the V, Ni, Co, Mo, As, Sb, Fe, Mn, Zn and Bi data for the Exxon oils enabled appropriate oils to be correlated with each other.