Magnetopneumography (MPG) can quantify the retention of magnetisable particles in the lung acquired, for instance, in welding work. MPG is non-invasive and is used in occupational health, industrial hygiene and lung physiology. Following a brief magnetisation, the remanent magnetic field is mapped with magnetometers outside the thorax. There is no unique analytical inverse solution to this class of magnetostatic problem, and various inverse methods have been proposed. In the present study, the influence of variations in size and shape of the lungs and chest, magnetic measurement noise, positional noise and spatial dust distribution are investigated in five inverse methods. The mean value of the field map, calibrated against a lung phantom, is the commonly used method. Lung and chest size influence the mean value method solutions strongly. Correction for chest size reduces these errors, but bias errors and sensitivity to the deposition pattern remains a problem. A multipolar expansion, including dipolar, quadrupolar and octopolar moments, yields best results overall, provided the single-to-noise ratio is sufficient. This inverse solution is unbiased, requires no calibration with phantom lung models and serves to minimise errors due to inter-individual differences in anatomy and to inhomogeneous retention of inhaled dust.