In-plant material supply has gained more interest since product diversity has increased with the growth of mass customization. In such production environment, mixed-model assembly lines are designed to support the production of a large scale of finished products obtained by the combination of alternative variants of components that ensure a specific functionality on the finished product. The availability of the required variant at each stage of the assembly process is critical for the production processes that operate in a just in time (JIT) environment as it’s the case for some first tier suppliers in the automotive industry. To ensure the availability of the required parts at each stage of the assembly process, several line feeding modes can be used. Their relative performance depends on several factors like the components’ diversity and physical features to name a few. The impact of these parameters on the performance of the line feeding modes is not well known and this issue remains weakly represented in the literature. Using an empirical approach, this study aims to analyse the performance of three line feeding modes (line stocking, kitting and sequencing) according to a set of key parameters. The performance of each line feeding mode is measured as a periodic total cost that includes preparation, picking for the assembly, transportation and storage costs.