Many public authorities rely on the use of non-chemical weed control methods, due to stringent restrictions on herbicide use in urban areas. However, these methods usually require more repeated treatments than chemical weed management, resulting in increased costs of weed management. In order to investigate the efficacy of four non-chemical weed control methods and glyphosate treatment, experiments were carried out on traffic islands in the growing seasons 2005 and 2006. Three trial sites were each divided into six treatment areas, which were either treated with glyphosate, flame, steam, hot air/flame, hot water or left untreated. The treatments were carried out at regular, predetermined intervals throughout the growing season in 2004, whereas in 2005 and 2006 how many treatments that were required to keep weed cover below a predetermined acceptance level of 2% were investigated. Percentage weed cover was measured every second week using a 75 cm × 75 cm quadrat divided into 100 squares. On the control areas, a rapid increase in weed cover was observed, whereas weed cover could be kept below 2% by 2–7 treatments per year, depending on control method. On average, the following numbers of treatments per year were required: glyphosate 2.5, hot water 3, flames 5, hot air/flames 5.5 and steam 5.5 treatments. The results demonstrate that the weed control should be adjusted to the prescribed quality for the traffic islands by regularly assessing the need for weed control. They also show that tailored treatments can reduce the number of required non-chemical treatments per year.