Abstract: Antibiotics are one of the emerging pollutants of concern in the world. Fast and efficient analytical methods for potential analysis of these pollutants are required, in order to control water quality and assure the safety of its use as a source of drinking water. However, there are few comprehensive methods to test multiple types of antibiotics in surface water. An online solid-phase extraction (SPE) of large-volume injection coupled with the high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) method was used for the simultaneous determination of 35 antibiotics in surface water, including 9 sulfonamides, 16 quinolones, 4 macrolides, 4 tetracyclines and 2 other classes. According to compound retention in SPE column and pH condition experiments, moxifloxacin and danofloxacin were confirmed as alkaline injection (0.1% ammonia solution), while the remaining 33 antibiotics were acidic injection (0.1% formic acid solution). The analytes were detected by multiple reaction monitoring (MRM) mode and were quantified by an internal standard method. The correlation coefficients of 35 antibiotics were greater than 0.995 with the calibration concentration range from 1−500ng/L. The method limits of detection (MLD) were low, ranging between 0.1−5.2ng/L. At 3 spiked concentrations of 5ng/L, 50ng/L, and 200ng/L, the relative standard deviation (RSD) of matrix labeling (n=6) were 1.65%−12.3%, 0.2%−10.4% and 0.05%−9.92%, respectively. The recovery rate of target compounds spike in the water samples were 62.1%−93.1%, 65.4%−127% and 65.9%−124%, respectively. The developed method only required a small sample volume (10mL) and simple pretreatment. When the acidic and basic injection volume were both 3mL, the total sample run time was only 30min including sample uptake, injection, online preconcentration, and detection. The developed method was applied to the analysis of 35 antibiotics in 21 surface waters, of which the detection rate of lincomycin reached 90.5%, and the highest content of ofloxacin reached 98.2ng/L. Compared with the conventional offline detection method, the detection results of the two methods were consistent, but the online detection method was simple, sensitive, accurate and fast, and capable of the real-time detection of water samples, making it suitable for residual analysis of trace antibiotics in surface water.