Sows suffering from clinical signs of disease (e.g. lameness, wounds and shoulder ulcers) are often involuntarily culled, affecting the farmer's economy and the welfare of the animals. In order to investigate the interrelationships between clinical signs of individual pregnant group-housed sows, we performed an explanatory factor analysis to identify factors describing the patterns of variation of clinical signs. Moreover, we investigated how these emerging factors affected the probability of a sow to be either (i) euthanized, (ii) suddenly dead, (iii) sent to slaughter due to clinical signs of disease such as claw lesions or wounds or (iv) involuntarily culled (representing a pool of sows that were either euthanized, dead or sent to slaughter due to disease). Data from 2.989 pregnant sows in group-housing systems from 33 sow herds were included in the study. A thorough clinical examination was performed for each sow by using a protocol including 16 different clinical signs. Farmers recorded all cullings and deaths and the reasons for these actions in a 3-month period after the clinical examination. Among the observed sows, 4.2% were involuntarily culled during the 3-month period. From the explanatory factor analysis, we identified three factors describing the underlying structure of the 16 clinical variables. We interpreted the factors as 'pressure marks', 'wounds' and 'lameness' Logistic analyses were performed to investigate the effect of the three factors and the parity number of each sow on the four outcomes: (i) euthanized, (ii) suddenly dead, (iii) sent to slaughter due to clinical signs of disease and (iv) involuntarily culled. The analyses showed that 'lameness' significantly increased the risk of sows to be involuntarily culled (P 5 0.016) or sent to slaughter due to clinical signs of disease (P 5 0.026). Lameness is generally considered to be an important welfare problem in sows, which could explain the increased risk seen in this study. By contrast, 'pressure marks' and 'wounds' did not have any significant effect on the four outcomes (P . 0.05).