To establish an accurate, reliable and stable method for determination of total arsenic in food.Methods Fish, seaweed (GBW 10023-GSB-14), rice (GBW 10010) and chicken (GBW 10018) were selected as food samples, which were pretreated by wet digestion, microwave digestion and dry ashing respectively, and determined by hydride generation atomic fluorescence spectrometry. Results The total arsenic results of dry ashing were 3.18±0.052,7.01±0.063,0.104±0.002 and 0.115±0.004 mg/kg which were in the range of reference values. The recovery rates were 93.7%-98.5%, and the RSD was below 2.31%. The results of wet digestion were 3.12±0.041,6.93±0.072,0.103±0.003 and 0.112±0.003 mg/kg which were in the reference range. The recovery rates were 89.2%-97.5%, and the RSD was below 1.92%. The results of fish and seaweed with microwave digestion were 1.74±0.032 and 15.40±0.096 mg/kg, which were less than 60% of the reference value, but the results of rice and chicken were 0.102±0.001,0.114±0.005 mg/kg, both in the reference range. The recovery rates were 96.3%-97.5%, and the RSD was below 2.51%. Wet digestion could be applied to most of the food samples for its flexibility in digestion temperature, time, type and dosage of acid. Microwave digestion consumed less time but large amount of acid, which could be applied to the relatively simple samples. Dry ashing method was suitable for samples of high volatile temperature, high oil content and complex arsenic structure because of its time-consuming.Conclusion Wet digestion and dry ashing could be applied as the pre-treatment of most food samples. Microwave digestion method has its limitations, only suitable for samples of arsenic with relatively simple morphology.