A numerical method has been developed recently to solve the Navier-Stokes equations for laminar compressible flow around delta wings. A large-scale Navier-Stokes solution on a mesh of 129 × 49 × 65 points for transonic flow Mx = 0·85, α = 10° and Rex = 2·38 × 106 around a 65° swept delta wing with round leading edge is presented and compared with a correspondingly large-scale Euler solution. The viscous results reveal the presence of primary, secondary, and even tertiary vortices. The starting location of the primary vortex is seen to be quite different in the two solutions. In the viscous solution it starts at the wing apex, but in the Euler results it starts about one quarter chord downstream. The secondary reparations are also different, due to the up-lifting of the boundary layer in the viscous results, but to a cross-flow shock in the Euler computation. Comparison with experiment shows that the interaction between the primary and secondary vortices in the Navier-Stokes computation is obtained correctly and that these results are a more realistic simulation than the one given by the Euler equations.