Quantification of the uncertainties associated with environmental geochemical prediction as a function of sample size is always a concern when performing specifically an Environmental Risk Assessment (ERA). While the study of determination of sample size in geochemical field is limited, there is a definite need for the development of a methodology together with a protocol which addresses geochemical sampling (sample sizing and representative) analyses in geochemical predictions. Based on Central Limited Theory, this article proposed a statistical methodology for determining sample size and applied this methodology using the Vaal River and West Wits tailing dams in South African as a case study. It also discussed the factors such as confidence interval, acceptable sampling error etc. that could influence sample size estimation and recommended a tradeoff strategy to reduce the sample size for economic reason.