variability within litter (BWvar) highly affect the pre-weaning survival. Further, the identification of low birth weight (LBW) piglets often relies on arbitrary statistical classifications. We established optimal cut-off values for BW and BWvar to predict pre-weaning piglet survival. BW, BWvar and survival data were obtained from FBN and Agroscope experimental pig facilities. The FBN data (2012 to 2021) for the German Landrace breed consisted of records from 28,242 total number of piglets born (TNB) i.e. born dead or alive from 752 sows. The Agroscope data (2004 to 2022) for the Swiss Large White consisted of 43,159 piglet records from 980 sows. The cut-off values for BW and BWvar to predict piglet survival from birth to weaning were estimated by the Receiver Operating Characteristic (ROC) curves analysis using the cutpointr package of R (v 4.0.3). Postnatal piglet mortality was highest within the first 3 days. From the piglets born alive, about 85.2 and 83.7% survived day 3 at FBN and Agroscope, respectively. Respective survival rates at weaning were 81.1 and 78.6%. Overall, the piglets below a BW cut-off value of 1.18 kg (i.e. FBN=1.17 kg and Agroscope=1.19 kg) had a lower survival probability at birth than those piglets>1.18 kg, and can be considered as LBW piglets. With a cut-off value of 0.278 g (AUC=0.55) the BWvar had a lower prediction accuracy than BW (AUC=0.66). Moreover, a BW cut-off value of 1.22 g (AUC=0.74) would increase the probability of piglets to survive until weaning. Although facilities were different in terms of genotypes and environments, highly similar cut-off values were estimated for identification of LBW with smaller survival probability for both facilities. The results also suggest that prediction of piglet survival with only BW is more informative than only with BWvar. The identified LWB piglets may benefit from extra support measures such as e.g. supplemental milk.
