Searching in a linguistically annotated treebank is a principal task that requires a sophisticated tool. Netgraph has been designed to perform the searching with maximum comfort and minimum requirements on its users.
Although it has been developed primarily for the Prague Dependency Treebank 2.0, it can be used with other treebanks too, both dependency and constituent-structure types, as long as the treebank is transformed to a suitable format. In this paper, we present Netgraph query language and on many examples show how it can be used to search for frequent linguistic phenomena.