This brief presents a complete methodology for evaluating achievable performances in applications of active noise control aiming robust broadband attenuation. A comparison between different control architectures, the influence of the number of available actuators and sensors on attainable performances, and the impact of considering the perturbation as measured or not are all discussed. First, a generic experimental device, specifically designed to study active noise attenuation, is identified as a multi-input multi-output system. Then, robust multiobjective control is considered, using a Hinfty criterion, and recasts as Hinfty optimization under constraints. Appropriate indicators are proposed for comparison purposes. Finally, quantified results are presented and validated experimentally. Intrinsic limitations introduced by non minimum phase zeros are illustrated and discussed.