An Escherichia coli arsRp::luc-based biosensor was constructed to measure the bioavailability of arsenic (As) in soil. In previous induction experiments, it produced a linear response (R (2) = 0.96, P < 0.01) to As from 0.05 to 5 μmol/L after a 2-h incubation. Then, both chemical sequential extraction, Community Bureau of Reference recommended sequential extraction procedures (BCR-SEPs) and E. coli biosensor, were employed to assess the impact of different long-term fertilization regimes containing N, NP, NPK, M (manure), and NPK + M treatments on the bioavailability of arsenic (As) in soil. Per the BCR-SEPs analysis, the application of M and M + NPK led to a significant (P < 0.01) increase of exchangeable As (2-7 times and 2-5 times, respectively) and reducible As (1.5-2.5 times and 1.5-2.3 times, respectively) compared with the no fertilization treated soil (CK). In addition, direct contact assay of E. coli biosensor with soil particles also supported that bioavailable As in manure-fertilized (M and M + NPK) soil was significantly higher (P < 0.01) than that in CK soil (7 and 9 times, respectively). Organic carbon may be the major factor governing the increase of bioavailable As. More significantly, E. coli biosensor-determined As was only 18.46-85.17 % of exchangeable As and 20.68-90.1 % of reducible As based on BCR-SEPs. In conclusion, NKP fertilization was recommended as a more suitable regime in As-polluted soil especially with high As concentration, and this E. coli arsRp::luc-based biosensor was a more realistic approach in assessing the bioavailability of As in soil since it would not overrate the risk of As to the environment.