The objective of this ACEA funded study was to determine the effect of different pedestrian autonomous emergency braking (P-AEB) systems on the collision speeds of real world pedestrian accidents originating from three different accident databases. The precrash phases of real world passenger car to pedestrian frontal accidents from the in-depth accident databases were investigated using different pre-crash simulation tools. Collision parameters were compared between the original real-world cases and cases with treatment conditions. For treatment simulations, the car was equipped with a virtual generic P-AEB system, triggered at a time to collision (TTC)≤ 1 s. The range of the generic sensor was 80 m and the opening angle was varied between 60°, 90° and 120°. For the braking system, two different brake gradients (24.5 m/s3 and 35 m/s3) were modelled with different decelerations of 0.8 g and 1.1 g. Accidents from the Austrian in-depth accident database CEDATU (n=50), the German GIDAS (n=1084) and Swedish V_PAD (n=68) were used for the baseline. The effect of using different data samples was compared to the effect of assuming different generic AEB system parameters. The best performing P-AEB system (120°, innovative brake system) avoided 42% of the CEDATU cases, while the baseline P-AEB system (60°, standard brake system) avoided 18%. The best performing AEB System was able to avoid 79.4% of the V_PAD sample. The baseline P-AEB avoided in V_PAD at least 66.2% compared to GIDAS with 39.5%. The lower the mean collision speed of the sample, the higher was the benefit of the P-AEB system, as a higher percentage of cases can be avoided. The study shows that system parameters and the selection of accidents can greatly affect the outcome in prospective traffic safety analyses. As a significant reduction of collision speeds was seen in all three data sources, the study highlights the need for a combined vehicle safety assessment instead of a separate evaluation of active and passive pedestrian safety measures.