At present, device-to-device (D2D) communication, due to its high reliability and low latency, plays a decisive role in the field of vehicular communication. However, how to efficiently allocate spectrum resources is still a huge challenge. In this paper, a heuristic algorithm is proposed to deal with access mode selection and resource management problem, which aims to maximize the sum transmission rate of the overall system while ensuring the quality of service (QoS) of D2D-enabled vehicular communication by considering the reliability of the link. In the heuristic algorithm, we firstly consider a new scenario that D2D-based vehicular users (denoted D-VUEs) autonomously and reliably reuse the resources of cellular vehicular users (denoted C-VUEs) and take some reasonable selection for C-VUEs and D-VUEs to determine the resource allocation for vehicle-to-vehicle (V2V) link and vehicle-to-infrastructure (V2I) link. Then, we consider the mode selection for resource allocation to avoid interference. In order to deal with the problem that channel state information (CSI) caused by mobility is difficult to obtain, we present the spectrum resource management model to achieve power control and spectrum sharing based on the channel model with large-scale slow fading and small-scale statistical characteristic of fast fading. The simulation results show that the algorithm obtains better performance and outperforms the existing scheme.