It is a challenging task to determine the dynamic vehicle attitude by using single-frequency single-epoch multi-antenna Global Navigation Satellite System (GNSS). In urban environment, the number of visible satellites drops sharply due to the occlusion of trees and tall buildings, hence it is difficult to obtain the high-precision attitude of vehicles only through multi-antenna GNSS. GNSS tight combination algorithm selects the same reference satellite between different systems, and can effectively increase the number of observation equations after eliminating the inter-system bias, to improve the attitude accuracy of vehicles in complex environments. Compared with the loose combination algorithm, which selects reference satellites separately between different systems, the tight combination algorithm can further improve the locatable performance when there are fewer satellites. Dynamic vehicle experiments were carried out in an open environment and complex environment respectively, using GPS/BDS/GALILEO three-system single-frequency and single-epoch positioning mode. The results show that the tight combination algorithm and the loose combination algorithm have the same accuracy when there are enough visible satellites in the open environment. In complex environment with a cut-off elevation angle of 40 degrees, the percentage of pitch angle error, yaw angle error and roll angle error within 2 degrees increased by 6.1%, 8.07% and 13.43% respectively , and the ambiguity fixed rate was increased by 14.78.