The　fatigue　life　of　concrete　is　affected　by　many　interwoven　factors　whose　effect　is　nonlinear.　Because　of　its　unique　self-learning　ability　and　strong　generalization　capability,　a　neural　network　model　is　proposed　to　predict　concrete　behavior　in　tensile　fatigue.　Firstly,　the　average　relative　impact　value　was　constructed　to　analyze　the　importance　of　parameters　affecting　fatigue　life,　such　as　the　maximum　stress　level　Smax,　stress　ratio　R,　failure　probability　P,　and　static　strength　f.　Then,　using　the　backpropagation　neural　network　improved　by　Bayesian　regularization,　S-N　curves　were　obtained　for　the　combinations　of　R=0,1,　0,2,　0,5;　f=5,　6,　7MPa;　P=5%,　50%,　95%.　Finally,　the　tensile　fatigue　results　obtained　from　different　testing　conditions　were　compared　for　compatibility.　Besides　utilizing　the　valuable　fatigue　test　data　scattered　in　the　literature,　insights　gained　from　this　work　could　provide　a　reference　for　subsequent　fatigue　test　program　design　and　fatigue　evaluation.　©　IABSE　Congress　Nanjing　2022　-　Bridges　and　Structures:　Connection,　Integration　and　Harmonisation,　Report.　All　rights　reserved.