Using the Lawrence Livermore National Laboratory (LLNL) torso phantom to calibrate a lung counting system can lead to the conclusion that three large area (i.e. >70 mm diameter) Ge detectors will outperform a four-detector array and provide a lower MDA as a four-detector array of large area Ge detectors covers a significant portion of inactive tissue (i.e. non-lung tissue). The lungs of the LLNL phantom, which are approximately 10 cm too short compared with real lungs, also suggests that a two-detector array could be used under limited circumstances. When tested with modified lungs that are more human-like, it was found that the four-detector array showed the best counting efficiency and the lowest MDA. Fortunately, these findings indicate that, although the LLNL phantom's lungs are too short, there is no adverse impact on the calibration of a lung counter.