Crating sows in farrowing systems greatly restricts their normal behaviour, which is usually justified by the assumption that piglet mortality is higher with loose-housed sows. Based on experiments showing that this is not the case, farrowing crates were banned in Switzerland in 1997. Since then, many farms have introduced loose farrowing systems, enabling a comparison of piglet mortality in farrowing systems with and without crates based on a large sample size. Data of a sow-recording scheme (UFA2000) were analysed using generalised linear mixed-effects models with an underlying Poisson distribution. In 2002 and 2003, the average total piglet mortality on 173 farms (n = 18,824 litters) with loose farrowing systems amounted to 1.40 piglets per litter and did not differ from that of 482 farms (n = 44,837 litters) with crates (1.42 piglets per litter). Nevertheless, the number of crushed piglets was significantly higher in pens with loose-housed sows (0.62 versus 0.52 piglets per litter), whereas the number of piglets that died for other reasons was significantly higher in crates (0.78 versus 0.89 piglets per litter). Total piglet mortality was influenced by litter size at birth, age of the sow and season. Consequently, evaluation of the reproductive data of commercial farms shows that no more piglet losses occur in loose farrowing pens, common nowadays in Switzerland, than in farrowing pens with crates, and that litter size at birth is the main influence on piglet losses.