L-moment procedures are applied to develop regional rainfall annual maxima (maximum 24-h rainfall occurring annually), considering complex physiographic and hydroclimatological features and weather processes as attributes influencing regional homogeneity. Capability of L-moment procedures in breaking down large areas into coherent and homogeneous sub-regions is investigated. The study area of 238,630 km<Superscript>2</Superscript> is located in southwestern Iran, representing 154 rainfall recording gauges of rather complex orographic features. According to the results and considering the extreme value nature of the analysis, i.e., rainfall annual maxima, it was possible to establish four homogeneous sub-regions, representing areas of 80,490 km<Superscript>2</Superscript>, 75,880 km<Superscript>2</Superscript>, 31,810 km<Superscript>2</Superscript> and 50,450 km<Superscript>2</Superscript>. The L-moment procedures successfully evaluated the homogeneity analyses and assignment of appropriate regional probability distribution functions (PDFs) to individual sub-regions. Flood potential maps and associated rainfall depth-area information were provided for individual sub-regions. The proposed heterogeneity measures by Hosking and Wallis (Water Resour Res 29(2):271–289, <CitationRef CitationID="CR20">1993</CitationRef>), known as H <Subscript>1</Subscript>, H <Subscript>2</Subscript> and H <Subscript>3</Subscript> tests were further investigated in this research. The results of this research confirmed findings of Hosking and Wallis (Water Resour Res 29(2):271–289, <CitationRef CitationID="CR20">1993</CitationRef>), i.e., strong capability of H <Subscript>1</Subscript> as a heterogeneity measure, and lack of capability of H <Subscript>2</Subscript> and H <Subscript>3</Subscript> measures for distinguishing between homogenous and heterogeneous regions. Copyright Springer Science+Business Media B.V. 2010
