A multiobjective bilevel programming is provided for seeking nondominated and sparse solutions to multiobjective optimization problems. By solving Pareto (efficient) solutions to the multiobjective bilevel programming, Pareto (efficient) solutions with sparse features which are restricted to a given set are given for multiobjective optimization. The solutions not only deal with large-scale multi-objective optimization problems effectively, but also reflect known preferences of the decision maker. In particular, the upper level objective functions involve a non-Lipschitz term. We establish M-, C- and S-stationary conditions for the non-Lipschitz multiobjective bilevel programming under the appropriate qualifications. Finally, a comparative experiment is given to illustrate that the multiobjective bilevel model can present some useful nondominated sparse information.