Many interlaboratory test programmes (ITPs) on rubber test methods, have been carried out during the last 10 years. ITPs have been organised within ISO TC 45 and also in several countries such as USA, UK and Sweden. Most of these ITPs have shown that the repeatability and the reproducibility are poor for many rubber test methods. In an attempt to do more than just determine the poor precision, we chose four methods and decided to study them to identify the factors giving poor precision, thereby improving the methods. This third part contains the work done on tensile tests. All of the investigated factors are influencing the results. However, using five test pieces instead of three and having a good calibration status is shown to be very important to improve precision in tensile testing.