Abstarct: In this research, at the first step a new isogeometrical level set topology optimization is introduced. In previous studies, the level set function is approximated by b-splines basis functions and the control net is updated during the optimization process. Since control points are not basically on the level set surface, a discrepancy between zero level of function that represents boundaries of the structure, and zero level of its control nets appeared. In this research, the idea is solving level set equation (LSE) over the parameter space of b-spline basis functions defined in the IGA model. Afterwards, the updated level set function is mapped into the physical space. The level set function is approximated over a grid defined in parameter space which is constantly a unit square. By doing so, the control net of the IGA model and the grid for solving LSE are separated. Two well-known radial basis functions (RBF) and reaction–diffusion (RD) baxsed methods are employed to solve the LSE. The proposed method is applied to minimize the mean compliance when a certain amount of material is used and also for weight minimization subject to local stress constraints. Several numerical examples are presented to demonstrate performance and accuracy of the proposed method. For the second step, this study intends to optimize the topology of structures at macro and micro scales simultaneously by using isogeometric analysis and the reaction-diffusion level set method. First, the geometry of structure on the macro and micro scales is modeled by the isogeometric analysis, then the level set function is defined on a grid of parameter space which is independent of the control net. Subsequently, at each optimization iteration, the equilibrium and homogenization equations are solved by isogeometric analysis. Sensitivity analysis at microscopic and macroscopic scales are performed to calculate the velocity at boundaries and the values of the level set function at both scales are updated by the reaction-diffusion equation. Finally, several examples with different geometry and boundary conditions are provided to show the performance and efficiency of the method. Also, the obtained results in comparison with the ones in the literature in terms of topology and final objective function value show good agreement.