The paper deals with the problem of finding a tandem scattered subsequence of maximum length (LTS) for a given character sequence. A sequence is referred to as tandem if it can be split into two identical sequences. An efficient algorithm for the LTS problem is presented and is shown to have O(n 2) computational complexity and linear memory complexity with respect to the length n of the analysed sequence. A conjecture is put forward and discussed, stating that the complexity of the given algorithm may not be easily improved. Finally, the potential application of the solution to the LTS problem in approximate tandem substring matching in DNA sequences is discussed.