Graphs are promising abstraction of complex structured and semi-structured data. Graph mining techniques extract, analyze and summarize significant and useful information from the graph databases. Finding frequent subgraph from graph database is an essence of graph mining. Sometimes the mined subgraphs are large in numbers, posing difficulty in selecting significant subgraph. Every frequent subgraph is not always significant from the application perspective. This paper proposes an innovative concept to extract significant subgraphs. The authors' method does this in two stages. In the first stage, frequent subgraphs are identified using frequency threshold, which is an input parameter.