River water and wastewater treatment plant (WWTP) effluents from metropolitan Taipei, Taiwan were tested for the presence of the pollutants estrone (E1), estriol (E3), 17beta-estradiol (E2), and 17alpha-ethinylestradiol (EE2) using a new methodology that involves high-flow solid-phase extraction and liquid chromatography/tandem mass spectrometry. The method was also used to investigate the removal of the analytes by conventional drinking water treatment processes. Without adjusting the pH, we extracted 1-L samples with PolarPlus C18 Speedisks under a flow rate exceeding 100 mL/min, in which six samples could be done simultaneously using an extraction station. The adsorbent was washed with 40% methanol/60% water and then eluted by 50% methanol/50% dichloromethane. The eluate was concentrated until almost dry and was reconstituted by 20 microL of methanol. Quantitation was done by LC-MS/MS-negative electrospray ionization in the selected reaction monitoring mode with isotope-dilution techniques. The mobile phase was 10 mM N-methylmorpholine aqueous solution/acetonitrile with gradient elution. Mean recoveries of spiked Milli-Q water were 65-79% and precisions were within 2-20% of the tested concentrations (5.0-200 ng/L). The method was validated with spiked upstream river water; precisions were most within 10% of the tested concentrations (10-100 ng/L) with most RSDs<10%. LODs of the environmental matrixes were 0.78-7.65 ng/L. A pre-filtration step before solid-phase extraction may significantly influence the measurement of E1 and EE2 concentrations; disk overloading by water matrix may also impact analyte recoveries along with ion suppression. In the Taipei water study, the four steroid estrogens were detected in river samples (ca. 15 ng/L for E2 and EE2 and 35-45 ng/L for E1 and E3). Average levels of 19-26 ng/L for E1, E2, and EE2 were detected in most wastewater effluents, while only a single effluent sample contained E3. The higher level in the river was likely caused by the discharge of untreated human and farming waste into the water. In the drinking water treatment simulations, coagulation removed 20-50% of the estrogens. An increased dose of aluminum sulfate did not improve the performance. Despite the reactive phenolic moiety in the analytes, the steroids were decreased only 20-44% of the initial concentrations in pre- or post-chlorination. Rapid filtration, with crushed anthracite playing a major role, took out more than 84% of the estrogens. Except for E3, the whole procedure successfully removed most of the estrogens even if the initial concentration reached levels as high as 500 ng/L.