Neural　networks　(NN)　are　applied　to　the　tracking　control　of　a　three-link　manipulator　attached　to　an　autonomous　underwater　vehicle　(AUV).　Lyapunov　design　is　employed　to　obtain　the　NN　based　robust　controller.　The　interaction　between　the　AUV　and　the　manipulator　is　considered.　Nonlinearity　in　the　plant　is　compensated　by　NN　based　identification.　To　illustrate　the　validity　of　the　proposed　controller,　numerical　simulation　is　performed　and　the　comparison　between　the　NN　based　controller　and　a　conventional　proportional-derivative　(PD)　controller　is　conducted.