Skilled　welders　can　estimate　and　control　the　weld　penetration　based　on　weld　pool　observation.　This　implies　that　an　advanced　control　system　could　be　developed　to　control　the　penetration　by　emulating　the　decision　making　process　of　the　human　welder.　In　this　paper　a　nonlinear　dynamic　model　is　established　to　correlate　the　process　inputs　(welding　current　and　traveling　speed)　and　weld　penetration　in　Gas　Tungsten　Arc　Welding　(GTAW).　An　innovative　3D　vision　sensing　system　capable　of　measuring　the　weld　pool　characteristic　parameters　in　real-time　is　utilized.　Dynamic　experiments　are　conducted　under　various　welding　conditions.　Dynamic　linear　model　is　first　constructed　and　the　results　are　analyzed.　The　linear　model　is　then　improved　by　incorporating　a　nonlinear　operating　point　modeled　by　Adaptive　Neuro　Fuzzy　Inference　System　(ANFIS).　It　is　found　that　the　penetration　state　can　be　better　modeled　by　the　proposed　ANFIS　model.　©　(2013)　Trans　Tech　Publications,　Switzerland.