During the first 5-month period of this 10-month study, arterial cannulations were performed using the standard technique based on palpation of pulsation or anatomic landmarks in 38 children. During the second 5 months, cannulations were performed with a NIRVIS in 29 children. The NIRVIS has a small NIR light source to transilluminate the puncture site. The light is projected on a display located above the puncture site, with blood vessels then visible as dark lines on a white background of the tissue. Patients were under general inhalation anesthesia with sevoflurane. The radial or ulnar artery at the wrist was the preferred site for the cannulations, performed with an over-the-needle technique (Abbocath 22-gauge or 24-gauge; Abbott, Chicago, Ill). If cannulation did not succeed at the wrist, the brachial or femoral artery was cannulated without the use of the NIRVIS, which is unable to visualize either of those arteries. The primary outcome measure was the total time to successful cannulation. Secondary outcomes were the time to the first flashback of blood, indicating that the artery was penetrated, and success at first attempt and number of punctures. Patient characteristics were comparable between the 2 groups. In 32 of the 39 patients in the NIRVIS group, the arteries were seen with the device. The median time to successful cannulation was 547 seconds without, and 464 seconds with, the use of the NIRVIS, not a statistically significant difference. The median times to the first flashback of blood were 171 and 219 seconds, respectively. Adjusting for confounders (weight for age, age, complicating factors, gender, anesthesiologist, mean arterial pressure, use of catecholamines) did not alter the results but did show that age and weight for age were predictors for a shorter time to successful cannulation and shorter time to first flashback of blood. The first attempt was successful in 7 of 38 patients without the NIRVIS and in 12 of 39 with the NIRVIS. This was not a statistically significant difference, however, after adjusting for confounders. Age was a predictor of success at first attempt with an odds ratio of 1.12. Three punctures were needed with the NIRVIS compared with 6 without the device. The success rate for cannulations in the wrist was 62% for the NIRVIS group compared with 50% for the non-NIRVIS group. This meant that the other 15 and 19 patients, respectively, required femoral or brachial lines. Although the use of NIRVIS did not substantially improve time and success rate of arterial cannulation in young children, it did offer advantages for success at first attempt and number of punctures. The results indicate that the insertion of the cannula, not localizing the artery, is the primary difficulty in performing arterial cannulation in small children.