A study was conducted to examine whether the New Zealand Code of practice for temporary traffic management guidelines for the implementation of temporary speed limits (TSL) result in driver speeds that match safe travelling speeds. Site approach and site entry speed data was collected at eight sites around Wellington, New Zealand, where TSLs were in place. Four TSLs (100 to 70km/h, 100 to 50km/h, 100 to 30km/k and 50 to 30km/h) and two visibility conditions (continuous and non-continuous) were used. Surveys were sent to 100 drivers at each site asking about their risk perceptions and attitudes to speeding and road works. While drivers did reduce speed from the approach to entry of a road works site, the reduced speeds, both the mean speeds and 85th percentile, were higher than the TSL. These findings indicate that improvements are needed regarding site design and TSL estimation to reduce the accident risk. Driver' subjective risk perceptions, either site specific or general, were not related to their site entry speeds. Drivers also tended to underestimate the speed at which they would drive through a site. The survey results suggest that drivers' subjective perceptions do not influence their objective behaviours. Recommendations are made for improving the setting of TSLs based on estimates of driver speeds.