Abstract: An analytical method was established for the determination of trace α-amanitin in the urine of patients suffering from mushroom poisoning by online solid phase extraction-liquid chromatography-tandem mass spectrometry (online SPE-LC-MS/MS). The sample was protein precipitated with formic acid acidified acetonitrile-methanol (5:1, v/v). Reversed-phase liquid-liquid microextraction was used to remove the organic solvent from the sample extract. The toxin was purified by online SPE using an ODS micro column (5 mm×2.1 mm, 5 μm), and separated on an XBridge^TM BEH C₁₈ column (150 mm×3.0 mm, 2.5 μm). Finally, the toxin was measured by MS/MS in the negative electrospray ionization (ESI^-) mode. Multiple reaction monitoring (MRM) was used, and the conditions were m/z 917.4>205.1 (quantitative ion transition) and m/z 917.4>257.1. Collision energy for both transitions was 55 eV. A fast valve-switching technique with a quantitative loop was used as an interface between the online SPE and LC-MS/MS modules. The two modules were independent, neither the mobile phase nor the pressure would interfere with each other, thus ensuring the stability of the system. Precise purification by the online system could effectively eliminate the matrix effects in the subsequent MS detection. Weak matrix suppression effects were found, with results of 88.7%-96.5%. The linear range of α-amanitin in urine was 0.1-50 μg/L with a correlation coefficient (r²) of 0.9983. The limit of detection (LOD) and limit of quantification (LOQ) in the sample matrix were 0.03 μg/L and 0.1 μg/L, respectively. The average recoveries at three spiked levels (0.1, 2.0 and 20 μg/L) were 84.3%-91.7% with relative standard deviations (RSDs) of 3.8%-7.2%. The accuracy and precision were evaluated using quality control samples with toxin contents of 0.1 μg/L (LOQ), 0.2 μg/L (2-fold LOQ), 2.0 μg/L (medium level), and 20 μg/L (high level). The calculated average intra-day accuracy was 85.1%-96.0% with the precision of 4.1%-7.8%. The inter-day accuracy was 82.9%-94.8% with the precision of 5.0%-9.5%. The specificity of the method was verified by negative samples derived from patients who suffered only gastroenteritis poisoning, without hepatotoxic symptoms. α-Amanitin was found in urine samples from nine mushroom poisoning patients with hepatotoxic symptoms. The sampling time ranged from 19 h to 92 h. The toxin contents were 0.11-53.1 μg/L. For patients with a high intake of poisonous mushrooms, the toxin content was 53.1 μg/L in a patient's urine sampled 19 h after accidental consumption and 0.19 μg/L in another patient's urine sampled 92 h after poisoning. The content of α-amanitin was only 0.53 μg/L in the urine sample obtained 23 h after consumption for a patient with low intake and 0.11 μg/L in the urine sampled from another patient 40 h after poisoning. Amatoxins can metabolize rapidly in vivo. The laboratory identification of amatoxin poisoning requires a method for trace-level analysis in the biological matrix. It is proved that this method is simple, accurate and sensitive by the application to the analysis of actual samples. The protein precipitation and reversed-phase liquid-liquid microextraction steps are fast and simple. Hence, they can be used as a rapid and effective pre-treatment method for online SPE-LC-MS/MS analysis of water-soluble toxins in biomaterial matrix. Highly sensitive analysis of α-amanitin in urine can be obtained using a precise purification technology via online SPE in this study. The problem of qualitative confirmation of the toxin at trace levels (0.03 μg/L) after poisoning can be solved. The laboratory identification time for amatoxin poisoning in some patients exceeds 90 h. The developed analytical method at trace level (0.1 μg/L of LOQ) can provide reliable technical support for establishing the dose-response relationship of α-amanitin in vivo. It can satisfy for the determination of trace α-amanitin in urine samples from patients with hepatotoxic mushroom poisoning.