This report describes a liquid chromatography (LC)-tandem mass spectrometry (MS) multiresidue method for determining traces of 30 base/neutral/acid herbicides and fungicides in bovine whole milk. Four milliliters of milk was spiked with the analytes and two surrogate analytes (SAs) and then diluted with 35 mL of a water/methanol solution (50:50, v/v). This mixture was passed through a 0.5-g Carbograph 4 (TM) cartridge. After washings, analytes were re-extracted by back-flushing the cartridge with 1.5 mL of methanol followed by 6 mL of methylene chloride/methanol (80:20, v/v), 50 mmol/L formic acid. After partial solvent removal down to about 0.1, 0.15 mL of I mmol/L formic acid aqueous solution and an internal standard (IS) were added. After filtration, 50 mu L of the final extract was then introduced into the LC analytical column. During the chromatographic run, the MS system was operated in both positive and negative ion modes. MS data acquisition was performed in the multi-reaction monitoring mode, selecting two precursor ion > product ion transitions for each target compound, except for pentachlorophenol. On analyzing six milk samples from different sources, absolute recovery of the analytes and the two SAs ranged between 78% and 104% with RSDs not larger than 13%. The accuracy of the method at three different spike levels was assessed by adding the two SAs to analyte-containing milk samples and varied between 82% and 120% with RSDs not larger than 11%. Limits of quantification were estimated to range between 0.008 and 1.4 mu g/L. Compared to the Carbograph 4 cartridge, one filled with a N-vinylpyrrolidone-m-divinylbenzene co-polymer (Oasis HLB (TM)) sorbent was much less efficient in recovering several of the acidic herbicides considered and, in addition, its relative final extract produced a severe negative matrix effect that drastically weakened ion signal intensities of several non acidic analytes.