This work presents the analysis of investigating methods concerning similarity of text documents in the area of the Polish law, basing on known designating similarity measures of texts among other things: Jaccard, Cosine, Sørensen-Dice measures and the VSM (Vector Space the Model) model with additional use of semantic network relations from Polish WordNET. There was proposed investigating method of legal texts semantic similarity consideration of their specificity. There was presented comparison of similarity investigation results for the designated legal documents corpus in case of the analysis without usage of semantic network WordNET lexical relations and with their participation.