Adhesion measurements on poly(dimethyl) siloxane samples were performed, for the first time, with flat glass probes under controlled tilt angle and the results were compared with measurements from spherical probes of two different radii. Experiments weremade on both flat and patterned samples with structure diameters of 4.7 mu m and heights of 0.82 mu m and 1.95 mu m, respectively. Pull-off forces measured with spherical probes showed the usual preload dependence and were independent of misalignment angle. On the other hand, pull-off forces measured with aligned flat probes were preload-independent, but dropped significantly and became preload-dependent with increasing misalignment. This effect was more pronounced for structured samples, where a misalignment by 0.2 degrees resulted in a drop of adhesion by more than 30%. The comparison indicates that measurements from spherical probes underestimate adhesive forces for structured surfaces if compared with aligned flat probes. Finally, we propose a simple model which allows the prediction of angle-dependent plateau values of pull-off forces for measurements with flat probes on flat samples.