In this letter, we jointly design relay antenna tilting with user scheduling so as to enhance the sum rate performance of a two-hop relay system, where the relay is equipped with a large-scale antenna array and performs zero-forcing processing. Building the fundamental of the joint design, a tight and tractable sum rate approximation is first derived by employing random matrix theory. Then the relay antenna downtilt and the number of active user pairs are jointly optimized to maximize the approximate sum rate. It is noted that the proposed scheme is independent of instantaneous channel state information. Therefore, it enjoys very low implementation complexity while improving the system performance.