A multilevel logistic regression model was used to identify and quantify the effects of factors affecting the mortality of pigs being transported to slaughter under commercial conditions. A survey of 739 journeys to 37 slaughterhouses in five eu countries was carried out, and information potentially related to the welfare and mortality of the pigs and the number of injuries were recorded. The average temperature during the journey, its duration, the average loading time per pig, the recorded injuries, fasting before transport, and the interaction between fasting and journey duration were used in the final model. The average mortality was 0·11 per cent and the average proportion of injured pigs was 0·36 per cent, and these figures were significantly correlated (P<0·001). In 29 per cent of the journeys the pigs were not fasted before being loaded, which doubled the risk of mortality irrespective of whether the pigs were injured or not. The risk of mortality increased with average temperature. In journeys with fasted pigs that did not have any recorded injury, average temperature was more important than the duration of the journey. The risk of mortality increased as the average time taken to load them decreased, and the risk was highest when the pigs were not fasted and when injuries were recorded. Other factors such as the country, loading density, availability of drinking water and type of ventilation did not affect the risk of mortality.