Extreme programming (XP) is the most well known agile software development method. Many experience reports have been published in recent years. Successful XP adoptions have however been criticized for the lack of concrete data. While some exist, the studies are often difficult to compare due to different settings and the varying level of XP adoption. This paper reports the first results (concrete data from 2/5 releases) from a controlled extreme programming case study. Four software engineers were acquired to implement a system in a tight delivery schedule of eight weeks. Development environment was close to the agile home ground. A comparison of the collected data from the first two releases is provided. Analysis shows that while the first release is a learning effort for all stakeholders, the second release shows clear improvement in all regards, e.g., estimation accuracy is improved by 26%, productivity was increased by 12 locs/hour and yet the post-release defect rate remained low, i.e., 2.1 defects/KLoc.