Crack data can be collected using manual or automatic surveys. Traditionally, manual methods are used, and they are still the most common. Changing into automatic systems will enhance the efficiency of data collection as well as the objectivity. In this study the repeatability of an automatic crack data collection system using video images was evaluated. Ten repetitive measurements were made on a 10-km-long road section. Cracking was measured using six laser cameras attached to the same vehicle. The results from the two methods were compared. Simultaneously, a simplified manual windshield survey was conducted by three different persons, and the repeatability obtained was used for comparison with the repeatability established using the image and laser method, respectively. The correlation between repeated measurements using the two automatic systems was high, while the repeatability for the manual, subjective method was low. Suitable measures for crack characterization are discussed.