In　regional　water　management　systems,　various　uncertainties　may　be　derived　from　random　feature　of　resource　conditions　and　natural　processes,　errors　in　estimated　modeling　parameters,　as　well　as　imprecision　or　fuzziness　human-induced.　This　leads　to　difficulties　in　formulating　and　solving　the　resulting　regional　water　management　problems.　In　this　study.　a　robust　multistage　interval-stochastic　programming　(RISP)　method　is　developed　for　dealing　with　vague　and　random　information　in　regional　water　management　systems.　The　decision　variables　are　useful　for　justifying　and/or　adjusting　the　decision　schemes　for　agricultural　activities　through　the　incorporation　of　their　implicit　knowledge　on　regional　water　management.　Different　policies　for　agricultural　water　supply　have　been　analyzed.　The　results　can　help　to　identify　desired　water-allocation　schemes　for　agricultural　sustainable　development　that　the　prerequisite　water　demand　for　supporting　crops＇　survival　can　be　guaranteed　when　the　water　resource　is　scarce.　(C)　2011　Elsevier　B.V.　All　rights　reserved.