Experiments and full resist simulations of contact patterns using both infinitely thin masks (2D) and 3-dimensional mask topography (3D) were performed to examine the quality of prediction by simulation. Experimental data were acquired by CD-SEM measurements of contact patterns in resist which were generated using a 193 nm scanner with a numerical aperture of 0.75, circular illumination (σ=0.5), and an attenuated phase shifting mask with 6% transmission. Analysis of the data is performed in terms of dose to size, process window, mask error enhancement factor (MEEF), and printed critical dimension (CD) in resist. Furthermore, an error analysis is performed with respect to mask CD, illumination source, dose and focus error. For the same contact size in resist a parabola like dependence of the mask contact length on contact width was found by experiment and simulation. Fair agreement between 2D and 3D simulation was obtained above 180 nm mask CD whereas a strong difference was observed below this region. Especially the location of the minimum at around 140 nm mask CD can be reasonably described only by 3D simulation. Thus, the prediction of accurate mask biases and process windows in the lower mask CD region is only possible by 3D simulation. Simple corrections of the 3D effect like the consideration of a mask CD offset or dose offset fail. Apart from that, 2D simulation in conjunction with a well calibrated resist model is sufficient for delivering reliable predictions for process window, MEEF, and CD.