In this paper a modified double smoothing bandwidth selector,MDS , based on a new criterion, which combines the plug-in and the double smoothing ideas, is proposed. A self-complete iterative double smoothing rule (_IDS ) is introduced as a pilot method. The asymptotic properties of both_IDS and_MDS are investigated. It is shown thath MDS performs asymptotically very well. Moreover, it is asymptotically negatively correlated with h ASE , the minimizer of the averaged squared error. The asymptotic performances of_MDS and of the iterative plug-in method,_IPL (Gasser et al., 1991) are compared. A comparative simulation study is carried out to show the practical perfor- mance of_MDS and related methods. It is shown that_MDS seems to be the best in the practice. Finite sample negative correlations between the chosen bandwidth selectors and h ASE are also studied.