Themodelingandpredictionofsuspendedsedimentinariverarekeyelementsinglobalwaterrecoursesandenvironmentpolicyandmanagement.Inthepresentstudy,anAdaptiveNeuro-FuzzyInferenceSystemmodeltrainedwiththeLevenberg-Marquardtlearningalgorithmisconsideredfortimeseriesmodelingofsuspendedsedimentconcentrationinariver.ThemodelistrainedandvalidatedusingdailyriverdischargeandsuspendedsedimentconcentrationdatafromtheSchuylkillRiverintheUnitedStates.TheresultsoftheproposedmethodareevaluatedandcomparedwithsimilarnetworkstrainedwiththecommonHybridandBack-Propagationalgorithms,whicharewidelyusedintheliteratureforpredictionofsuspendedsedimentconcentration.ObtainedresultsdemonstratethatmodelstrainedwiththeHybridandLevenberg-Marquardtalgorithmsarecomparableintermsofpredictionaccuracy.However,thenetworkstrainedwiththeLevenberg-MarquardtalgorithmperformbetterthanthosetrainedwiththeHybridapproach.