A rain gauge network should serve general as well as specific purposes such as water supply, hydropowergeneration, flood forecasting, irrigation, and flood control. The level of accuracy a network can achieve depends onthe number and locations of gauges in the network. In this study, a rain gauge network was designed for Hormozganprovince in the south of Iran. Monthly rainfall totals from 124 rain gauge stations in the period from 2000 to 2009were used. This province can be logically divided to four regions using the De Martonne aridity index. Kagan’sapproach was used to relocate the rain gauge network to obtain the optimal design. In this statistical method, thecorrelations were classified based on distance. Exponential models were fitted to the average correlations againstmean distances in all regions. The number of gauges and the distance between gauges were computed to satisfy userrequirements. The results showed that Hajiabad had the minimum value for distance (125 km) and Bandar Lengehhad the maximum value for distance (588 km). Spatial variation of rainfall in Hajiabad was greater than for otherstations. The results indicate that 40, 50, 20, and 55 stations were adequate to represent rainfall with 15% averageerror in the regions of Bandar Lengeh, Bandar Abbas, Hajiabad and Minab, respectively.