In this paper, we take the long steel production line as research background, and investigate the charge batching and batch sequencing problem on the continuous-caster and the rolling machine. Different from the previous research in which the casting and rolling stages are disjoined, in this study the batching and batch sequencing requirements on both stages are considered simultaneously, as well as the requirement on balancing the workloads on different finishing machines downstream. A novel mixed integer programming (MIP) model is developed. Because of the NP-hardness of the problem and the huge size of MIP model, along with practical requirement for industry application, we propose an improved scatter search (SS) algorithm to solve the problem. In this algorithm, the search space is reduced by some optimal properties and the variable neighborhood search strategy is introduced such that the diversification and the complementary advantage of difference neighborhoods are combined and the merits of hybrid algorithms are exploited. The computational results on practical data demonstrate the efficiency of the improved scatter search algorithm.