Although computerized provider order entry (CPOE) systems are being more widely implemented and appear to reduce medication errors, little data exists on the effectiveness of specific CPOE systems. This study evaluated the implementation of two widely used off-the-shelf CPOE systems (with limited decision support) and found that both resulted in significant reductions in serious medication errors. The article also details types of new errors induced by CPOE systems, which, while common, were generally not clinically significant. As the evidence base around implementation of CPOE systems remains relatively small, studies like this that evaluate the real-world performance of information technology are increasingly important.