Lean Thinking (Womack and Jones) improves quality, cost and delivery through the relentless elimination of the wastes. For example, the exemplar of Lean, the Toyota Production system, focuses on improvement through the continual elimination of seven categorised wastes (Ohno). Time compression (Stalk and Hout) is a key weapon in attacking this waste. By compressing time, quality issues are revealed driving improvement and cost reduction. When the value stream is mapped (Hines and Rich), inventory is usually the largest target for time compression. As inventory is removed from process buffers, process time variability has a deleterious impact on throughput. Lean production fundamentals (housekeeping, standardisation & waste elimination) focuses on reducing this variability. In this paper we investigate a production scenario with high variability - a multi-class and re-entrant system (Dai). For this system this paper makes two contributions: - 1. Simulates the relationship between buffer size and throughput performance. 2. Investigates the potential for improvement through Lean Thinking There is an intricate relationship between loss of throughput and allowed inventory buffer sizes along the production flow. In simple cases, as for instance in single product unidirectional flows, there will normally be a strong reciprocal trade-off, whenever buffers are squeezed towards zero, but in production systems where the workflow can be characterised as multi-class and re-entrant, there is much more complexity, requiring simulation to gain understanding. Process buffer capacity is strongly dependent on the capacity, variability, overall workflow and scheduling priority. This suggests that a systems view is necessary in order to understand the trade-offs between loss of throughput and lack/excess of buffer capacity in such systems. The findings are put into a long-term perspective based on lean thinking recommendations. Based on systems thinking (Towill and Naim), four generic sequential steps to Lean Supply and Distribution have been developed within the Lean Paradigm (Simons and Kiff) - Control, Time, Centralisation and Structure. Control focuses on removing variability from the information and physical flows through standardisation of processes. Empirical evidence from automotive after-sales demonstrated that simulation could identify the areas of variability with greatest leverage on cost and quality (Simons, Kiff and Cheiux). Hence, as well as understanding the trade-offs in the current scenario, this paper generically investigates the opportunities for re-balancing the throughput and buffer capacity of the system.