The expanding use of surfactants for proteome sample preparations has prompted the need to systematically optimize the application and removal of these MS-deleterious agents prior to proteome measurements. Here we compare four different detergent clean-up methods (Trichloroacetic acid (TCA) precipitation, Chloroform/Methanol/Water (CMW) extraction, commercial detergent removal spin column method (DRS) and filter-aided sample preparation(FASP)) with respect to varying amounts of protein biomass in the samples, and provide efficiency benchmarks with respect to protein, peptide, and spectral identifications for each method. Our results show that for protein limited samples, FASP outperforms the other three clean-up methods, while at high protein amount all the methods are comparable. This information was used in a dual strategy of comparing molecular weight based fractionated and unfractionated lysates from three increasingly complex samples (Escherichia coli, a five microbial isolate mixture, and a natural microbial community groundwater sample), which were all lysed with SDS and cleaned up using FASP. The two approaches complemented each other by enhancing the number of protein identifications by 8%-25% across the three samples and provided broad pathway coverage.