Aiming at the online multi-sourcing multi-attribute reverse auctions (OMSMARA) with existence of interactive negotiation in reality, and taking into account the uncertainties of different aspects faced by buyers and sellers, this paper uses bi-level programming theory and fuzzy theory to research the decision-making questions in OMSMARA with bilateral negotiation under uncertainties. Firstly, based on the description of problems and the appropriate assumptions, a new fuzzy mixed integer bi-level programming (FMIBLP) model with interaction and negotiation is established, which is accurately conversed through augmented fuzzy maximin decision method. Then, considering the characteristics of the problem model and the advantages of PSO algorithm, a bi-level distribution iterative algorithm (PSO-BLDI) based on modified PSO is proposed. Subsequently, the feasibility of the model and the effectiveness of the algorithm are demonstrated by the numerical example and comparative analysis. Finally, the sensitivity analysis is used to study the influence of relevant parameter change on the model results, which further demonstrates the rationality of the model and the effectiveness of the decision-making method.