Generally, when a prototype is scaled down to a scale model, all parameters of the scale model should be designed to satisfy the similitude criteria. For a base-isolated structure with rubber bearings, however, it is technically difficult to obtain smallsized rubber bearings with the first shape factor satisfying the similitude criteria. This distortion affects the similitude ratio of the vertical stiffness of the bearings and, thus, influences the similitude ratio of the horizontal response of the structure because of the overturning effect. To solve this problem, a coordinative similitude method is proposed for scale model testing of structures with rubber bearings by adjusting configuration of the isolators to match the similitude ratio of the overturning stiffness of the isolation layer. A full-scale and two 1/3 scaled bearings were designed and tested, and their similitude relations were evaluated. Test results of the model bearings were extrapolated to the OpenSEES model of a conventional model building and a coordinative building designed based on the conventional similitude method and the coordinative similitude method, respectively. A comparison between seismic responses of two model buildings against the prototype building indicates that the coordinative similitude method has significantly larger accuracy than the traditional similitude method. K E Y W O R D S first shape factor, overturning effect, rubber bearings, scale model testing, similitude criteria, similitude method 1 | INTRODUCTION Scale models are widely used for investigating the seismic behavior of different types of structures. [1,2] For long-span or super high-rise structures, geometric scale factors can be as large as 100 or larger. [3-5] The scale model testing relies on satisfying the similitude criteria between the model structure and the prototype structure. [6-8] When a base-isolated structure with rubber bearings are scaled down for a shaking table test, however, it is difficult to obtain scale model rubber bearings with all of their parameters satisfying the similitude criteria. The geometric scale factor of the superstructure needs to be designed small enough to accommodate the dimension and payload of the shaking table, and as a result, the dimension of the scale model of the rubber bearings are designed very small in dimension. Table 1 lists the scale factors and the geometric details of some scale model rubber bearings used for shaking table tests in literature. [9-16] It can be noted that scale model rubber bearings need to be designed with diameters smaller than 150 mm in some cases. However, in these cases, the individual rubber layer thickness cannot be scaled down correctly. As stated in Kalpakidis, [14] "this is necessary when bearings are produced in an industrial environment: steel shims and rubber layers are available in specific minimum thickness that allows for reliable manufacturing." The deviation in rubber layer thickness reduces the first shape factor of the bearing, which is defined as the ratio of the effectiv...