Selection of bacterial consortia in the biotreatment tank affects its performance. Three approaches of constructing the bacterial consortia from a given sludge to treat a specific wastewater were compared in this paper. In the all-pollutant approach, all the pollutants in the wastewater were used in sludge enrichment and bacterial isolation. In the partial-pollutant approach, only some of the pollutants were used. In the separate-pollutant approach, the pollutants in the wastewater were divided into groups. Consortia obtained from each group of pollutants were combined to form the final consortium. No comparison among these three approaches has been studied so far. Two dyes, reactive blue 250 and reactive blue 19, served as our model system for evaluating the approaches and performance was quantified by the extent of decolorization. Results showed that the all-pollutant approach was better than the partial-pollutant approach or the separate-pollutant approach. This result could be rationalized by examining the interactions during bacterial consortium construction. In addition, a case study with nine dyes was conducted; it showed that decolorization was poor because not all the dyes were used in consortium construction.