A　novel　domain　decomposition　method　is　proposed　in　this　paper　to　solve　eigenvalue　problems.　Both　the　simple　and　multiple　eigenvalues　are　considered　in　algorithm　design　and　theoretical　analysis.　The　key　thought　is　to　transform　the　eigenvalue　problem　into　a　number　of　linear　boundary　value　problems　in　a　multilevel　space　sequence　and　some　small-scale　eigenvalue　problems　in　a　low-dimensional　correction　space.　Then　some　well-developed　domain　decomposition　methods　for　linear　boundary　value　problem　can　be　directly　used　to　resolve　eigenvalue　problems.　Through　rigorous　theoretical　analysis　and　numerical　experiments,　we　can　find　the　presented　novel　domain　decomposition　algorithm　for　eigenvalue　problems　can　derive　the　same　solving　efficiency　and　parallel　scalability　as　that　for　linear　boundary　value　problems.