Fluvial sediment transport studies have long underscored the difficulty in reliablyestimating transport model parameters, collecting accurate observations, and makingpredictions because of measurement error, natural variability, and conceptual modeluncertainty. Thus, there is a need to identify modeling frameworks that accommodate theserealities while incorporating functional relationships, providing probability-basedpredictions, and accommodating for conceptual model discrimination. Bayesian statisticalapproaches have been widely used in a number of disciplines to accomplish just this, yetapplications in sediment transport are few. In this paper we propose and demonstrate aBayesian statistical approach to a simple sediment transport problem as a means toovercome some of these challenges. This approach provides a means to rigorously estimatemodel parameter distributions, such as critical shear, given observations of sedimenttransport; provides probabilistically based predictions that are robust and easilyinterpretable; facilitates conceptual model discrimination; and incorporates expertjudgment into model inference and predictions. We demonstrate a simple unisize sedimenttransport model and test it against simulated observations for which the ‘‘true’’ modelparameters are known. Experimental flume observations were also used to assess theproposed model’s robustness. Results indicate that such a modeling approach is valid andpresents an opportunity for more complex models to be built in the Bayesian framework.