Economic lot-scheduling problem (ELSP) has been studied since the 1950's. ELSP deals with the scheduling of the production of several products on a single machine in a cyclical pattern. The machine can only produce one single product at a time, and there is a set-up cost and set-up time associated with each product. Researchers generally adopted two types of rounding off methods for the production frequency of products, namely, the nearest integer and power-of- Two approaches. Production frequency of products defines the number of times that such product being produced during the cycle. Therefore, different production frequency actually leads to different optimization results. For this reason, this paper proposes a modified hybrid genetic algorithm to deal with this problem. Numerical examples are used to test the performance of the new approach. Results demonstrate the significance of the production frequency to the optimization results.