This article presents comparisons between predictions, obtained during the course of this investigation, and recently produced measurements of the flow development through a square cross-sectioned U-bend of strong curvature, Rc/D = 0.65, that is either stationary or in orthogonal rotation. For the stationary case, four turbulence models have been tested; a high-Re k-epsilon model interfaced with the low-Re 1-equation model in the near-wall regions, a high-Re algebraic second-moment (ASM) closure with the low-Re 1-equation model in the near-wall regions, and two versions of a low-Re ASM model. The two low-Re ASM models return noticeably better predictions of the flow development. There is, however further scope for improvement, especially in the downstream section. Two rotating flow cases have been computed both with the axis of rotation parallel to the axis of bend curvature; one at a positive rotation number R(0) = Omega D/W-b of 0.2 and one at R(0) = -0.2. In the case of positive rotation, where the Coriolis and curvature forces reinforce each other, the flow predictions of the low-Re ASM are in very close agreement with the data. When the section is not well reproduced by the low-Re ASM model. More refined turbulence models are thus necessary when the curvature and Coriolis forces oppose each other.