The FemCap is a new silicone rubber barrier contraceptive shaped like a sailor's hat, with a dome that covers the cervix, a rim that fits into the fornices, and a brim that conforms to the vaginal walls around the cervix. It was designed to result in fewer dislodgments and less pressure on the urethra than the cervical cap and diaphragm, respectively, and to require less clinician time for fitting. This was a phase II/III, multicenter, randomized, open-label, parallel group study of 841 women at risk for pregnancy. A subset of 42 women at one site underwent colposcopy. Women were randomized to use the FemCap or Ortho All-Flex contraceptive diaphragm, both with 2% nonoxynol-9 spermicide, for 28 weeks. The objectives were to compare the two devices with regard to their safety and acceptability and to determine whether the probability of pregnancy among FemCap users was no worse than that of the diaphragm (meaning not more than 6 percentage points higher). The 6-month Kaplan-Meier cumulative unadjusted typical use pregnancy probabilities were 13.5% among FemCap users and 7.9% among diaphragm users. The adjusted risk of pregnancy among FemCap users was 1.96 times that among diaphragm users, with an upper 95% confidence limit of 3.01. Clinical equivalence (noninferiority) of the FemCap compared with the diaphragm, as defined in this study, would mean that the true risk of pregnancy among FemCap users was no more than 1.73 times the pregnancy risk of diaphragm users. Because the observed upper 95% confidence limit (and even the point estimate) exceeded 1.73, the probability of pregnancy among FemCap users, compared with that among diaphragm users, did not meet the definition of clinical equivalence used in this study. The FemCap was believed to be safe and was associated with significantly fewer urinary tract infections. More women reported problems with the FemCap with regard to insertion, dislodgement, and especially removal, although their general assessments were positive. The two devices were comparable with regard to safety and acceptability, but a 6-point difference in the true 6-month pregnancy probabilities of the two devices could not be ruled out. Further studies are needed to determine whether design modifications can simplify insertion and removal.