Small changes to software are sometimes likely to cause unpredicted and negative effects on other parts of the software, especially for object-oriented Java programs due to the complicated relations between their components. Software Change impact analysis (CIA) is a key technique for identifying these potential effects. Most existing CIA techniques focus on method level without considering various changes at different granularity levels. In this article, we propose a new CIA technique, HSMImpact, which is performed in three steps: firstly, hierarchical change sets are defined at different granularity levels; then a CIA process is promoted based on the Hierarchical slicing model (HSM) technology; finally Hierarchical impact sets (HIS) are computed from package-level to statementlevel, which show that precision can be improved at finer granularity level. Preliminary case studies demonstrate the effectiveness of our technique.