I have recently applied super-efficiency model to rank efficient DMU's using 'Benchmarking' package in R. However, I am unsure as to how we should interpret the results. Out of the total DMU's, 20 were found efficient under output oriented VRS model. Using super efficiency model I was able to distinguish between 7 of them only (whose efficiency was greater than 1). Efficiency of 10 DMU's turned out to be Inf while the efficiency of 3 DMU's remained at 1 (7+10+3=20). Therefore, I would like some suggestions as to how we should interpret these results. Should we drop the DMU with Inf value as outliers? If not, how should we rank them as well as the DMUs with efficiency 1?