The　long-term　deformations　of　mountain　tunnels,　which　attract　more　and　more　attentions,　are　closely　related　to　the　time-dependent　features　of　the　surrounding　rock　mass.　However,　it　is　not　easy　to　determine　an　appropriate　rheological　model　and　its　corresponding　parameters　for　a　certain　engineering　instance.　This　paper　presents　a　rheological　parameter　estimation　technique　by　using　error　backpropagation　neural　network　(BN)　and　genetic　algorithm　(GA).　The　application　of　the　proposed　technique　to　an　engineering　instance,　Ureshino　tunnel　line　I　on　Nagasaki　expressway,　is　expatiated　in　detailed.　The　stochastic　nature　of　the　proposed　technique　is　also　discussed　through　case　studies.　It　is　proved　that　the　proposed　technique　can　provide　the　engineer　with　an　optimal　estimation　of　the　rheological　parameters,　which　can　help　the　prediction　of　long-term　deformations　of　mountain　tunnels　in　the　future.　©　2008　Elsevier　Ltd.　All　rights　reserved.