Ozonation and subsequent post-treatments are increasingly implemented in wastewater treatment plants (WWTPs) for enhanced micropollutant abatement. While this technology is effective, micro pollutant oxidation leads to the formation of ozonation transformation products (OTPs). Target and suspect screening provide information about known parent compounds and known OTPs, but for a more comprehensive picture, non-target screening is needed. Here, sampling was conducted at a full-scale WWTP to investigate OTP formation at four ozone doses (2, 3, 4, and 5 mg/L, ranging from 0.3 to 1.0 gO(3)/gDOC) and subsequent changes during five post-treatment steps (i.e., sand filter, fixed bed bioreactor, moving bed bioreactor, and two granular activated carbon (GAC) filters, relatively fresh and pre loaded). Samples were measured with online solid-phase extraction coupled to liquid chromatography high-resolution tandem mass spectrometry (LC-HRMS/MS) using electrospray ionization (ESI) in positive and negative modes. Existing non-target screening workflows were adapted to (1) examine the formation of potential OTPs at four ozone doses and (2) compare the removal of OTPs among five post treatments. In (1), data processing included principal component analysis (PCA) and chemical knowledge on possible oxidation reactions to prioritize non-target features likely to be OTPs. Between 394 and 1328 unique potential OTPs were detected in positive ESI for the four ozone doses tested; between 12 and 324 unique potential OTPs were detected in negative ESI. At a specific ozone dose of 0.5 gO(3)/gDOC, 27 parent compounds were identified and were related to 69 non-target features selected as potential OTPs. Two OTPs were confirmed with reference standards (venlafaxine N-oxide and chlorothiazide); 34 other potential OTPs were in agreement with literature data and/or reaction mechanisms. In (2), hierarchical cluster analysis (HCA) was applied on profiles detected in positive ESI mode across the WWTP and revealed 11 relevant trends. OW removal was compared among the five post-treatments and 54 -83% of the non-target features that appeared after ozonation were removed, with the two GAC filters performing the best. Overall, these data analysis strategies for non-target screening provide a useful tool to understand the behavior of unknown features during ozonation and post-treatment and to prioritize certain non-targets for further identification. (C) 2018 Elsevier Ltd. All rights reserved.