Abstract: We develop a genetic algorithm (GA) for a real-life problem of scheduling of the golf-club-head injection process performed on a set of parallel machines. In our study, the orders for club heads are usually requested by sets with certain due dates and can be shipped only if they are complete. Items (club heads) are scheduled individually without pre-emption. The cost to be minimized consists of earliness and tardiness penalties around due dates. Earliness involves the inventory holding cost of each item completed before the due date, whereas tardiness penalties are incurred when a complete order is delivered after its due date. GA solutions are compared with earliest due date (EDD) solutions and with the schedules used in a real-world setting. The results indicate that the schedule provided by GA is superior to both EDD and the real-world schedules in terms of cost savings. Moreover, we provide evidence that the algorithm delivers optimal or near-optimal solutions for a large number of randomly generated problems.