Image　watermarking　capacity　research　is　to　study　how　much　information　can　be　hidden　in　an　image.　In　watermarking　schemes,　watermarking　can　be　viewed　as　a　form　of　communications.　Almost　all　previous　works　on　watermarking　capacity　are　based　on　information　theory,　using　Shannon　formula　to　calculate　the　capacity　of　watermarking.　This　paper　presents　a　blind　watermarking　algorithm　using　Hopfield　neural　network,　and　analyzes　watermarking　capacity　based　on　neural　network　for　the　first　time.　Result　shows　that　the　attraction　basin　of　associative　memory　decides　watermarking　capacity.　©　2004　IEEE.