The arrangement of error microphones for a vehicle active noise control (ANC) system is no trivial work, especially for heavy-duty trucks due to the dilemma resulted from the large volume of the cab and the limited number of microphones accepted by most manufacturers in the auto industry. Most ANC developers resort to trial and error, which is not only a heavy workload given the amount of interested working conditions to be test, but also does not guarantee to yield the optimum noise cancellation performance. In this paper, the authors proposed an error microphone selection methodology using a genetic-algorithm (GA) -based mechanism. The target vehicle was a heavy-duty truck with a six-piston diesel engine, and two scenarios were particularly interested, i.e. driver & copilot and driver & one passenger sleeping on the berth. We first arranged eight microphones at different locations in the cab, five on the headrests, two on the B pillars and one at the head position of the sleeping berth. These locations were selected based on our empirical experience, the geometrical feature of the cab and the target application scenario. With this layout, the engine-induced acoustic signals at the microphone positions along with the engine rotation rate under different working conditions (idling and constant speeds at different gears) were measured for subsequent analysis. Then, a GA-based numerical optimization targeting at reducing the low-order engine noise using three error microphones was conducted, yielding that one error microphone on the B pillar, one on the headrest and one at the end of the sleeping berth led to the optimum noise attenuation performance. Road tests validated the numerical result.