Attenuated total reflectance infrared microspectroscopy (ATR-IRMS) combined with multivariate analysis was used to study the efficiency of different cleaning protocols applied to remove membrane fouling after membrane emulsification. An organic nitrocellulose mixed esters (MCE) membrane was used to prepare oil-in-water (O/W) emulsions. After the emulsification process, the membranes were cleaned with Tween 20 solutions at different concentrations (2%, 3% and 4%) in backwash mode with N-2 pressure (150, 500 or 700 kPa). The efficiency of the membrane cleaning process was also assessed by water flux recovery and membrane surface characterization by ATR-IRMS in the mid-infrared region (4000-800 cm(-1)) The analysis of the raw spectra showed that the surface of fouled and cleaned membranes had mainly sunflower oil. Soft independent modeling of class analogy models (SIMCA) were created using the raw ATR-IRMS spectra to differentiate between new, fouled and cleaned membranes. Class projections of transformed ATR-IRMS spectra and interclass distances values showed clear differentiation between the clusters of different cleaning protocols tested, and using discriminating power IR bands sunflower oil was identified as the only foulant agent on the MCE membrane surface. This study shows the potential of applying ATR-IRMS combined with SIMCA models to study membrane fouling/cleaning of organic microfiltration membranes. (C) 2014 Elsevier Ltd. All rights reserved.