This thesis deals with DEA models (Data envelopment analysis), which are models for measuring the efficiency of peer decision making units (DMUs). First part focus on basic DEA models (CCR model, BCC model) and this part is extended by new CH-CCR model, which considering that the variables (inputs or outputs) are usually strongly correlated. In second part is included description of network-based models, which are afterwards applicated on data of FRS CR (Fire Rescue Service of the Czech Republic). Using this models is solved efficiency of FRS CR through the regions.