The counterfeit coin problem is well-known and truly interesting in Computer Science, Game theory, and also in Mathematics. In this problem the objective is to detect the fake coin(s) of identical appearance but different weight in minimum number of comparisons. The word counterfeit most frequently describes forgeries of currency or documents, but can also describe software, pharmaceuticals, clothing, and more recently, motorcycles and cars, especially when these result in patent or trademark infringement. Finding one fake coin among n coins is tricky enough and complex. The problem becomes rigorous when there are two fake coins, as the false coin pair may form several different combinations that make the problem particularly tricky and complex to solve. In this paper we have developed a new algorithm for solving two versions of the two counterfeit coins problem in O(log n) time, where n is the number of coins given. © 2014 IEEE.