Sorting is a fundamental operation in computer science. Sorting means rearranging data in some order, such as ascending, descending with numerical data or alphabetically with character data. There are many sorting algorithm that have been proposed to meet the particular application, among which is insertion sort. It is known that insertion sort runs faster when the list is ‘nearly’ sorted but it runs slow when the list is in reverse order. This paper shows a way to improve the performance of insertion sort technique by implementing the algorithm using a new approach of implementation. This new approach has compared with the original version of insertion sort algorithm and bubble sort and showed that the proposed approach performed better in the worst case scenario.