Objective To identify risk factors strongly associated with failed mid-cavity vacuum-assisted delivery (VAD) and construct a risk model that can be used to guide obstetricians in choosing between the standard delivery room or the operating theatre. Design Observational analytical predictive prospective monocentric study. Setting University Hospital of Besançon, France, 2010-2018. Population All successful (n=903) and failed (n=194) attempts at mid-cavity VAD among term singleton deliveries. Methods Bivariate and multivariate analyses with logistic regression were used to determine the effects of 15 potential factors on the risk of VAD failure. Main outcome measures Mode of delivery following an attempt at mid-cavity VAD. Results The score comprised seven variables significantly associated with failed mid-cavity VAD: head-perineum distance ≥51 mm (p<0.001), bulky caput succedaneum (p<0.001), maternal height ≤1.54 m (p<0.001), duration of expulsive efforts before vacuum <21 minutes or >30 minutes (p=0.0013), duration from five centimetres to complete cervical dilation ≥3 hours (p=0.0091), term ≥42 weeks of gestation (p=0.032), and occiput-posterior or occiput-transverse fetal head position (p=0.041). The model was reliable (Hosmer-Lemeshow test =8.5; p=0.39) and accurate (concordance index =0.74). The threshold for a transfer to the operating theatre was set at 16.3 %. Advanced preparation of caesarean section equipment and anticipated extension of epidural analgesia were significantly associated with a decrease in the delay between VAD failure and birth by caesarean section (p<0.001). Conclusions This model could help obstetricians to better assess the risk of failed mid-cavity VAD and to choose the appropriate place to perform it. Keywords Vacuum-assisted delivery; mid-cavity.