Purpose:　Progressive　bone　erosion　(BE)　is　a　prominent　manifestation　of　structural　damage　in　rheumatoid　arthritis　(RA).　This　study　aimed　to　construct　and　validate　a　BE　prediction　model　(BEPM)　for　RA　based　on　musculoskeletal　ultrasound　and　clinical　risk　factors.　Methods:　A　total　of　312　RA　patients　without　BE　were　consecutively　collected　and　followed　up　for　2 years,　who　were　divided　into　BE　group　and　non-bone　erosion　group,　as　confirmed　by　two　radiology　experts　based　on　at　least　two　imaging　techniques.　Relevant　clinical　information,　such　as　anti-cyclic　citrullinated　peptide　(ACCP)　antibody　and　disease　duration,　was　collected.　Musculoskeletal　ultrasound　examinations　were　performed　on　all　patients　upon　admission.　Univariate　and　binary　logistic　regression　analyses　were　conducted　to　identify　risk　factors　for　BE　and　to　establish　the　prediction　model.　The　calibration,　diagnostic　efficacy,　and　clinical　effectiveness　of　BEPM　were　evaluated　by　calibration　curve　plots,　receiver-operating　characteristic　curve,　and　decision　curve　analysis　(DCA).　Results:　Binary　logistic　regression　analysis　screened　four　variables,　including　synovial　hyperplasia,　synovial　blood　flow,　ACCP　+,　and　disease　duration,　into　the　prediction　model.　BEPM　exhibited　good　diagnostic　performance　in　both　the　training　and　validation　groups,　with　area　under　the　curve　values　of　0.949　(95%　CI　0.924–0.973)　and　0.965　(95%　CI　0.935–0.996),　respectively.　Using　0.376　as　the　optical　cutoff　value　for　BE,　the　sensitivity,　specificity,　and　Youden　index　of　the　model　in　the　training　group　were　86.0%,　90.4%,　and　76.4%,　respectively.　In　the　DCA　curves,　BEPM　indicated　overall　good　population　benefit.　Conclusion:　BEPM　demonstrates　high　diagnostic　performance　and　clinical　utility,　and　may　be　a　useful　clinical　model　for　predicting　BE　in　patients　with　RA.　(Table　presented.)　©　The　Author(s),　under　exclusive　licence　to　International　League　of　Associations　for　Rheumatology　(ILAR)　2024.