Two one-way eccentric, two-storey, one-by-one-bay reinforced concrete (RC) structures are pseudodynamically tested under unidirectional ground motions. Theoretical considerations about the effect of torsional coupling on modal periods and shapes agree with modal results of the test structure, considering member stiffness is equal to the secant stiffness to yielding in skew-symmetric bending. Modal periods of such an elastic structure are in fair agreement with effective periods inferred from the measured response at the beginning of a test of a thoroughly cracked structure and at the end of the test. A time-varying stiffness matrix and a non-proportional damping matrix fitted to the test results may be used to reproduce the measured response approximately by modal superposition and identify the role of the four time-varying modes. Flexible side columns sustained very large drift demands simultaneously in the two transverse directions and suffered significant but not heavy, damage at lap-splices. RC-jacketing of the flexible side columns practically eliminated the static eccentricity between the floor centres of twist and mass as well as the torsional response. Inelastic time-history analysis with point-hinge member models, using as elastic stiffness the secant stiffness to yielding and neglecting post-ultimate-strength cyclic degradation of resistance in members with plain bars and poor detailing, predicted fairly well the response until the peak displacements and member deformations occurred. After that, it underestimated displacement peaks and the lengthening of the apparent period and missed the gradual drifting of the response towards a permanent offset. 1071 Figure 3. Input motion, scaled to PGA of 0.1g: (a) applied record (after scaling time); (b) 5%-damped elastic spectrum of original record (before scaling time), compared to smooth target spectrum; and (c) evolutionary 5%-damped elastic spectra of applied record (after scaling time) in 5 s windows of original time (before scaling).