Our objective was to determine the factors involved in the development of shoulder dystocia in association with operative vaginal delivery. In this prospective study, patients who were candidates for operative vaginal delivery were randomized either to forceps (N 5 315) or vacuum with M-cup (N 5 322) and timed from initial placement of instrument to final delivery. Data were gathered prior to and after instrumental delivery. Statistics used were Pearson chi square, Fisher's exact, analysis of variance, and multiple logistic regression. There were a total of 21 patients with shoulder dystocia in both groups (3.3% incidence). Discriminant factors that did not meet significance included use of epidural analgesia (P 5 .12), station (P 5 .99), previous vaginal delivery (P 5 .99), fetal gender (P 5 .54), indication for operative vaginal delivery (P 5 .63), .45 degree rotation (P 5 .68), use of episiotomy (P 5 .62), maternal weight (P 5 .26), and maternal diabetes (P 5 .08). Nearly attaining significance in univariate analysis was randomization to vacuum (P 5 .052). Significant factors included gestational age (P 5 .03), time required to effect delivery (P 5 .007), and birthweight (P 5 .0001). When these factors were subjected to stepwise multiple logistic regression, three factors remained as significant associations with shoulder dystocia: randomization to vacuum (P 5 .04), time for delivery (P 5 .03), and birthweight (P 5 .0001). In this operative vaginal delivery trial, shoulder dystocia was strongly associated with large fetal size, longer time to delivery, and the use of vacuum for delivery.