Toyota's goal-chasing method has proved effective for dealing with practical-sized sequencing problems for mixed-model assembly lines in just-in-time (JIT) production systems. Miltenburg introduced the multi-level concept to JIT assembly line sequencing, and his remark about Toyota's neglect of product ‘usage’ smoothing (PUS) when determining the assembly-line product sequence, primarily gave impetus to this study. We note that for most JIT systems, PUS would only be of practical significance, if it helps to achieve sub-assemblies and parts usage smoothing (SUS). In this paper, we conduct a simulation experiment for two-level sequence scheduling for JIT assembly lines. Using the SUS objective value as the metric, the results show that in a comparatively higher percentage of cases (relative to single-level SUS objective scheduling), which can be reasonably considered as statistically significant, better sequences are produced, This phenomenon is found to apply to Toyota's SUS objective as well as Miltenburg's.