Mixed-μ analysis (hereafter also called worst-case analysis) techniques have become a powerful practice for supporting the control system design process on complex, multivariable, uncertain, time-varying and even non-linear systems. The motivation for the adaptation of such tools to space problems originated from the fact that large spacecraft or space structures exhibit low-frequency (quasi) undamped flexible modes leading to risks of interaction with high-accuracy attitude control. Although these problems have been routinely solved by control engineers for years, the large parametric uncertainty in the plant sometimes leads to destabilizing worst-case combinations that are not intuitive at all and detected late in the development process. Astrium has successfully made use of these techniques on a variety of space applications, providing the development teams with valuable insights into the intricacies of complex systems. The article more specifically addresses the case of the polar-orbiting weather spacecraft MetOp scheduled for launch in 2005. The analysis campaign successfully detected all worst-case combinations of parameters that could destabilize the system: most were already known, but the process also allowed to discover some non-intuitive combinations involving antenna flexibilities.