We investigate balancing U-type assembly lines under uncertainty. We formulate the robust problem and develop an optimization model to design lines that hedge against disruptions. Offering more alternatives to group the operations, U-type assembly layouts are shown to be more efficient than conventional straight lines. As a result, they have been widely investigated in literature. However, a great majority of the studies assume deterministic environments and ignore various sources of uncertainty, like variability in operation times. We address this research gap. To hedge against variations in operation times, we employ robust optimization that considers worst case situations. However, this approach could result in pessimist solutions. In order to avoid that, we will assume that only a subset of operation times will be assigned to their worst case values. To solve the model, we will make use of Benders Decomposition, however it may converge slowly for large instances. To be able to solve real life problems, we will propose approximate solution algorithm. The efficiency of the algorithm will be evaluated with some computational tests.