Records of length of productive life, from first farrowing to culling, of 16,464 Large White purebred sows from SUISAG were studied using survival analysis. The major aims of the study were to model the risk of culling within parity and to assess the influence of exterior traits, such as the number of teats or feet and leg scores, on culling. Culling was concentrated at the first day after each farrowing or at the first day after weaning. Weaning itself was mostly between 21 and 49 d after farrowing, with an average weaning age of 35 d. Because of the definition of culling date used, there was practically no risk of culling from these periods. The culling rates at different periods suggested a modeling of the baseline hazard function within parity instead of over the entire productive life of the animals. A piecewise Weibull function and a simple graphical method to validate its adequacy were proposed for sow longevity analysis. The risk of culling increased with older parities (P < 0.001) and with decreasing litter size at weaning (P < 0.001). The exterior traits analyzed (number of teats, and feet and leg scores, on a scale from 1 to 7) had a moderate effect on the risk of culling compared with other factors but were still influential on survival, productive life expectancy, and annual replacement rate. Sows with less than 13 good teats had 1.35 times greater risk of being culled than sows with more good teats (P < 0.05). Sows with an X-O rear leg score of 2 had 1.4 times greater risk of being culled than sows with an intermediate score of 4 (P < 0.05). Sows at the optimum score of 4 for the size of inner claws of the rear leg had 0.83 times less risk of being culled (P < 0.01) than sows with scores of 2 and 3. Furthermore, when a phenotypic index for feet and legs was used to group these variables, the effect was highly significant (P < 0.001). Therefore, a means to improve longevity is through phenotypic selection of replacement gilts based on exterior traits: gilts with 13 or less good teats or with extreme feet and leg scores should be culled. From a genetic point of view, sows with the best value in the current index for exterior traits had a lower risk of culling (P < 0.01), and therefore, it is possible to obtain a response for sow longevity via indirect selection for exterior traits. From 1999 to 2003, the trend has been to eliminate extreme animals on exterior traits. This may partly explain the improvement of sow length of productive life longevity from 560 d in 2000 to nearly 710 d in 2003 observed in the data set.