Background:　Albumin　is　believed　to　be　associated　with　the　prediction　of　various　cancers.　This　retrospective　cohort　study　aimed　to　explore　the　non-linear　relationship　between　albumin　level　and　overall　survival　(OS)　in　advanced　non-small　cell　lung　cancer　(NSCLC)　patients　with　anlotinib　therapy,　utilizing　the　Generalized　additive　model　(GAM)　approach.　Specifically,　we　investigated　the　potential　non-linear　associations　that　might　not　be　captured　by　conventional　linear　analyses.　Methods:　A　retrospective　cohort　of　211　patients　undergoing　anlotinib　treatment　for　advanced　NSCLC　was　included　in　this　study.　A　wide　range　of　albumin　levels　was　considered,　and　the　GAM　method　was　applied　to　account　for　potential　confounding　clinical　variables　and　unveil　the　non-linear　relationship　between　albumin　and　OS.　Results:　A　non-linear　relationship　with　inflection　points　of　40　g/L　and　48　g/L　was　detected　between　albumin　level　and　OS　after　adjusting　for　potential　confounders.　The　hazard　ratio　(HR)　of　the　left,　middle　and　right　of　the　inflection　points　were　0.95　(95　%　confidence　interval　[CI],　0.87　to　1.04,　p　=　0.2819,　n　=　71),　0.75　(95　%　CI,　0.64　to　0.89,　p　=　0.0007,　n　=　133)　and　2.79　(95　%　CI,　1.39　to　5.61,　p　=　0.0039,　n　=　7),　respectively.　The　findings　indicate　a　negative　correlation　between　albumin　level　and　OS　when　albumin　level　was　between　40　g/L　and　48　g/L.　For　every　unit　increase　in　albumin,　there　was　a　25　%　reduction　in　the　risk　of　death.　Subgroup　analysis　revealed　that　the　negative　relationship　was　enhanced　with　blood　urea　nitrogen　(BUN)　level　increase　and　diminished　with　D-dimer　increase.　Conclusions:　The　relationship　between　albumin　level　and　OS　was　non-linear.　Albumin　level　is　an　independent　prognostic　factor　for　OS.　In　addition,　BUN　level　and　D-dimer　level　could　modify　the　effect　of　albumin　level　on　the　risk　of　death　in　advanced　NSCLC　patients　treated　with　anlotinib.　©　2023