In a market environment where food safety problems still occur despite repeated prohibitions, food safety problems caused by veterinary drug residues and biological safety problems caused by the transfer of drug resistance have attracted much attention. Herein, a method based on a compound purification system coupled with direct analysis in real time-tandem mass spectrometry (DART-MS/MS) was established to determine 41 different types of veterinary drug residues in livestock and poultry products. First, a single-standard solution sampling method was used to optimize the selection of the best quasi-molecular ion, two daughter ions, and their cone-hole and collision voltages; qualitative and quantitative ion pairs are composed of a quasi-molecular ion and its corresponding daughter ion. The abundance ratios of the drug compounds in standard solutions of the solvent and matrix mixtures were then calculated according to the requirements of the European Union 2002/657 specification. DART-MS/MS was subsequently developed for the accurate characterization and quantitative analysis of the veterinary drugs. Finally, a composite purification pretreatment system was formed by combining the primary secondary amine (PSA) and octadecyl bonded silica gel (C18) of a QuEChERS technology with multiwalled carbon nanotubes (MWCNTs) to achieve the one-step purification of the drug compounds. The influence of the key parameters of the DART ion source on the determination of the drugs was investigated using the peak areas of the quantitative ions as the criterion. The optimum conditions were as follows: ion source temperature of 350 ℃, 12-Dip-it Samplers module, sample injection speed of 0.6 mm/s, and external vacuum pump pressure of -75 kPa. According to the differences in the dissociation constant (pKa) ranges of the 41 types of veterinary drug compounds and the characteristics of the sample matrixes, the extraction solvent, matrix-dispersing solvent, and purification method were optimized based on the recovery. The extraction solvent was 1.0% acetonitrile formate solution, and the pretreatment column included MWCNTs containing 50 mg of PSA and 50 mg of C18. The three chloramphenicol drugs showed a linear relationship in the ranges of 0.5-20 μg/L with correlation coefficients of 0.9995-0.9997,and the detection and quantification limits of three chloramphenicol drugs were 0.1 and 0.5 μg/kg, respectively. The 38 other drugs, including quinolones, sulfonamides, and nitro-imidazoles showed a linear relationship in the ranges of 2-200 μg/L with correlation coefficients of 0.9979-0.9999, and the detection and quantification limits of the 38 other drugs were 0.5 and 2.0 μg/kg, respectively. The recoveries of the 41 veterinary drugs at low, medium, and high spiked levels in chicken, pork, beef, and mutton samples were 80.0%-109.6%, with intra- and inter-day precisions of 0.3%-6.8% and 0.4%-7.0%, respectively. A total of 100 batches of animal meat (pork, chicken, beef, and mutton; 25 batches each) and known positive samples were simultaneously analyzed using the national standard method and the detection method established in this study. Sulfadiazine (89.2, 78.1, and 105.3 μg/kg) was detected in three batches of pork samples, and sarafloxacin (56.3, 102.0 μg/kg) was detected in two batches of chicken samples and no veterinary drugs were detected in the other samples; both methods yielded consistent results for known positive samples. The proposed method is rapid, simple, sensitive, environmentally friendly, and suitable for the simultaneous screening and detection of multiple veterinary drug residues in animal meat.