The EU project Welfare Quality (R) proposes an overall assessment system for animal welfare based on animal outcomes. The objective of this study was to test inter-observer reliability (IOR) when assessing lameness, fear and slipping and falling scores as parameters for monitoring the welfare of killing pigs during arrival at the slaughterhouse. Two Belgian and two Spanish slaughterhouses were visited by six to seven observers. Lameness, slipping and falling were assessed twice; during unloading and in the passageway to the lairage zone (lairage). Fear, which was assessed in the unloading area, was based on four indicators: reluctance to move, retreat attempts, turning back and vocalisations. Lameness had low-to-moderate IOR when observed in the passageway to lairage (r = 0.46), but the IOR was low during unloading (r = 0.25). IOR for slipping and falling was moderate to high (r = 0.71 and r = 0.50, respectively), when assessed in the unloading area, but low for observations in the passageway (r = 0.13). Fear indicators had only moderate or low IOR. Turning back was the measure with the highest IOR (r = 0.43) and retreat attempts had the lowest IOR (r = 0.25). Based on these results, we concluded that scoring lameness could be reliable when assessed from the unloading bay to lairage, whereas slipping and falling should be scored in the unloading area of the slaughterhouse. We suggest scoring a maximum of two measures of fear on the same animals at the unloading area, with the most reliable parameters being turning back and reluctance to move. The three indicators of animal welfare (lameness, fear and slipping and falling) should be measured in a way to reduce overtax of the observers in order to achieve accurate results.