Percutaneous endovascular aneurysm repair (PEVAR) has been shown to be feasible; however, technical success is variable, reported to be between 46.2% and 100%. The objective of this study was to quantify the learning curve of the PEVAR closure technique and identify predictors of closure failure. We reviewed patient- and procedure-related characteristics in 99 consecutive patients who underwent PEVAR over a 30-month period in a single academic institution. A suture-mediated closure device (Proglide or Prostar XL) was used. Forward stepwise logistic regression was used to investigate associations between the failure of the closure technique and a number of patient and operative characteristics. To ensure objective assessment of the learning curve, a time-dependent covariate measuring time in calendar quarters was introduced in the model. Poisson regression was used to model the trend of observed failure events of the percutaneous technique over time. Overall PEVAR technical success was 82%. Type of closure device (P<.35), patient's body mass index (P<.86), type of anesthesia (P<.95), femoral artery diameter (P<.09), femoral artery calcification (P<.56), and sheath size as measured in Fr (P<.17) did not correlate with closure failure rates. There was a strong trend for a decreasing number of failure events over time (P<.007). The average decrease in the odds of technical failure was 24% per calendar quarter. The predicted probability of closure failure decreased from 45% per patient at the time of the initiation of our PEVAR program to 5% per patient at the end of the 30-month period. There were two postoperative access-related complications that required surgical repair. Need for surgical cutdown in the event of closure failure prolonged the operative time by a mean of 45 minutes (P<.001). No groin infections were seen in the percutaneous group or the failed group. Technical failure can be reduced as the surgeon gains experience with the suture-mediated closure device utilized during PEVAR. Previous experience with the Proglide device does not seem to influence the learning curve.