This paper describes a series of vehicle emission tests on a port-fuel injected lean-burn engine, to determine the preferred aftertreatment configuration yielding the most efficient regeneration of a lean-NOx trap (LNT). Three configurations were tested: (A) single starter three-way catalyst (TWC) upstream of an underfloor LNT; (B) bifurcated system with short downpipes comprising parallel TWCs upstream of a single underfloor LNT (Y-pipe configuration); and (C) bifurcated system with extended downpipes. System ‘A’ exhibits satisfactory LNT regeneration behaviour, and is within the European Stage III limits after accelerated aging. Results for system ‘B’, with identical TWC and LNT formulations as the single system, show that this LNT cannot be adequately regenerated under standard purge conditions; even with a fresh trap. In this non-optimized bifurcated system, the AFR profile entering the LNT during the rich purge deviates markedly from that requested by the calibration. A change in TWC formulation to one with different oxygen storage characteristics demonstrates that this system can perform satisfactory in the bifurcated configuration. Further tests on system ‘C’ with: (1) uncoated TWC substrates; and (2) the original TWC formulation, indicate that LNT regeneration efficiency in a bifurcated system is dependent upon the TWC formulation in conjunction with oscillatory flow in the Y-pipe. Implications of these results are discussed in terms of future direct injection gasoline engine programs.