Two ultrasonic devices – flat (T1) and curved (T2) ultrasonic transducers – were developed to remove biofilms from opened and closed surfaces, respectively. The aim is to standardize biofilm removal for in situ sanitary control in the food industry. The biofilms studied in this work were model biofilms made with milk on stainless steel sheets. We have shown in a previous study that sonication could be employed to remove and resuspend biofilm consistently, with a good recovery rate, from opened surfaces. Plate counting was used to assess the efficiency of each treatment. A total removal of Escherichia coli and Staphylococcus aureus from model biofilms was obtained with T1: 10 s at 40 kHz. However, ultrasound applied with T2 (a patented curved transducer developed for closed surfaces: 10 s at 40 kHz) failed to completely remove these model biofilms: 30±7% and 66±10% for E. coli and S. aureus biofilms, respectively. In order to improve the biofilm removal from closed surfaces with T2, the effect of the application of ultrasound in combination with chelating agent preparations was investigated. The application of ultrasound with T2 in 0.05 mol EDTA or EGTA per litre dislodged the E. coli milk model biofilm, with 100±10% and 100±5% recovery yields, respectively. These results showed a synergism between ultrasonic waves and chelator preparations, i.e. the combination achieved three times the recovery rate of sonication alone (30%). However, when the same treatment was applied to the S. aureus milk model biofilm, the combined treatment with EDTA or EGTA did not significantly improve the recovery of the biofilm cells: 74±26% with EDTA at 0.025 mol/l and 41–47% with EGTA at 0.025 mol/l and 0.05 mol/l, respectively, compared with 66±10% for sonication alone. The combined treatment was in agreement with an industrial control, i.e. a good reproducible recovery of the biofilm in a few seconds (10 s) for E. coli milk biofilms but not for S. aureus biofilms.