In this research, an analysis of the efficiency of Croatian counties in tourism was conducted. The objectives of this analysis were: to explain the concept of efficiency in tourism, to critically analyze previous research, to conduct empirical research using Date Envelopment Analysis (DEA), to interpret the results obtained and to discuss research hypotheses. The BCC model of output-oriented variable yields was used. The number of permanent beds was used as an input in the calculation of relative efficiency, while the number of tourist arrivals and the number of tourist nights were used as outputs. With the help of DEA, the results were found that 6 counties are relatively efficient out of the observed 21 counties, so that 15 of them are relatively inefficient. Relatively efficient counties are: Dubrovnik-Neretva County, Istria County, Krapina-Zagorje County, Međimurje County, Virovitica-Podravina County and the City of Zagreb. Near the efficiency limit are: Split-Dalmatia County, Brod-Posavina County, Koprivnica-Križevci County and Primorje-Gorski Kotar County. Two hypotheses are set in the paper: Two hypotheses are set in the paper: H1. There is a difference in the efficiency of Croatian counties in tourism, and H2. Based on the AOMP method, it is possible to make recommendations to relatively inefficient Croatian counties in order to reach the efficiency limit. Both hypotheses were accepted. Each relatively efficient county can achieve relative efficiency by changing its outputs, with unchanged input.