In this study, we compared the performance of conventional sample preparation techniques used in mycotoxin analyses against automated on-line sample clean-up for the determination of deoxynivalenol (DON) and its conjugated derivative, deoxynivalenol-3-β-d-glucoside (D3G), in cereal grains. Blank wheat and barley samples were spiked with DON and D3G, extracted with a mixture of acetonitrile:water (84:16, v/v) and processed by one of the following: extract and shoot, MycoSep(®) 227 clean-up columns, MycoSep 227 with an additional acetonitrile elution step and centrifugal filtration, followed by analysis with liquid chromatography tandem mass spectrometry. Based on method performance characteristics and poor recoveries (<30%) obtained for the polar D3G with some techniques, the extract and shoot approach was chosen for the inter-laboratory method comparison study. Thus, the same spiked samples were analysed in parallel by another laboratory with an in-house validated on-line sample clean-up method, utilising TurboFlow™ chromatography coupled to high resolution mass spectrometry. Method validation was performed by determination of specificity, linearity, recovery, intra-day precision and the limits of detection and quantification. Matrix-matched linearity (R(2)>0.985) was established in the range of 100-1600 and 20-320μg/kg for DON and D3G, respectively. Average recoveries (%RSD) were acceptable with both methods for wheat and barley, ranging between 73% and 102% (3-12%) for DON and 72% and 98% (1-10%) for D3G. The benefit of using automated sample clean-up in comparison to extract and shoot is the ability to inject directly pure extracts into the mass spectrometer, offering faster analyses and improved sensitivity with minimum system maintenance.