This article proposes a technique combining a sizing optimizer with metamodelling for railcar structural optimization. The authors describe a two-level or hierarchical, interactive, and meta-model-based optimization (HIMO) approach. At the lower level, a sizing optimizer finds feasible and optimal solutions in terms of sizing variables (plate thickness in continuum structures). Performance constraints such as stress, displacement, and stability, are handled only at this level. At the upper level, a metamodel is built to fit all the optimal solutions found at the lower level and is optimized for topology design. This results in the number of topology design variables being much smaller than those used in many other topology optimization approaches. The authors describe how this HIMO approach was applied to two boxcar design projects, resulting in 18 per cent and 36 per cent weight savings and significant reductions in manufacturing cost and total cost. The authors conclude that both the pilot tests and the real applications demonstrate that the HIMO approach works for real layout optimization.