In initiating a chorionic villus sampling (CVS) protocol, we relied upon placental locations as the determinant in the choice of technique. An anterior or fundal location prompted a transabdominal (TA) CVS, while the transcervical (TC) approach was reserved for posterior placentas. A coaxial needle system was used for TA CVS (18-gauge, 15-cm guide needle and 20-gauge, 20-cm sampling needle), while TC CVS was accomplished with a 5.8-French, 27-cm polyethylene catheter. Between July 1988 and February 1989 our initial 118 procedures were performed for 115 consecutive pregnancies using this protocol. Testing indications and antenatal characteristics of the TA (n = 56) and TC (n = 63) groups were similar. One procedure failure occurred in the TC group, and a single aspiration was sufficient in 59% of the cases (TA, 30/55; TC, 39/63). In TC procedures an increased aggregate sample weight was observed as compared to TA cases (25.8 g vs. 16.9 g, respectively; P less than .001). This difference was not attributable to an increased number of placental aspirations in TC cases. One abnormal karyotype was observed (45X), and four pregnancy losses occurred (TC, 3; TA, 1). Using placental location to determine the choice of CVS technique appears to be feasible and may be associated with a lower failure rate during a facility's initial experience (when compared to reliance upon one technique alone). Trials comparing the safety of these two methods should consider placental location an independent variable before randomization.