The　design　of　soil　nail　walls　in　China　is　dictated　by　two　national　technical　specifications,　namely,　Technical　Specification　for　Retaining　and　Protection　of　Building　Foundation　Excavations　by　China　Academy　of　Building　Research　(CABR)　and　Specifications　for　Soil　Nailing　in　Foundation　Excavations　by　China　Association　for　Engineering　Construction　Standardization　(CECS).　Each　specification　proposes　a　method　for　computation　of　maximum　nail　loads　while　the　prediction　accuracy　of　the　method　is　still　not　yet　fully　assessed.　To　fill　the　gap,　this　study　evaluates　the　accuracies　of　the　two　methods　using　144　measured　nail　loads　collected　from　fully　instrumented　soil　nail　walls　in　China.　The　influences　of　three　wall　working　conditions　on　the　prediction　accuracy　of　the　methods　are　also　examined.　The　results　show　that　the　two　default　methods　overestimate　the　maximum　nail　loads　by　about　40%　on-average　and　the　spread　in　prediction　generally　ranges　from　70%　to　100%.　However,　the　prediction　accuracy　is　not　significantly　influenced　by　wall　working　conditions.　Simple　calibrations　with　two　or　three　empirical　constants　are　introduced　to　the　two　methods　to　achieve　great　improvements　in　prediction　accuracy.　The　calibrated　CABR　and　CECS　methods　are　demonstrated　to　be　unbiased　on　average　and　the　prediction　dispersions　are　reduced　to　about　50%.　©　2019　Elsevier　Ltd