A multi-analyte method for the liquid chromatography-tandem mass spectrometric determination of mycotoxins in crude grain extracts without clean-up has been applied to the analysis of spelt, rice and barley. Method performance characteristics were determined after spiking blank samples at multiple levels and were found to be comparable for all investigated matrices as regards linearity (linear calibration functions were obtained for all analyte/matrix combinations except for moniliformin), precision (coefficient of variations <6%) and sensitivity. Matrix-induced signal suppression/enhancement was studied in detail and varied significantly between the investigated matrices, as well as between individual samples (relative standard deviation was as high as 40% within three rice varieties) and individual toxins. It was concluded that a reliable quantitative analysis using matrix-matched calibration requires careful consideration of the model matrix, which should match the investigated samples as close as possible.