The comparison of mass standards in air generally requires air buoyancy corrections. To determine these corrections it is necessary to estimate the density of moist air inside the balance case during the weighing procedure. The relative combined standard uncertainty of the air density determination is about 10⁻⁴ when using the CIPM-81/91 formula. The aim of our present work is to reduce the air density uncertainty using other methods and the purpose of this paper is to compare the performance of the following three methods: (a) application of the CIPM-81/91 formula (absolute determination); (b) direct determination using air buoyancy artefacts having the same surface area but a large volume difference (absolute determination); (c) refractometry, exploiting the high correlation between air density and air index of refraction using a novel refractometer well adapted to our requirements (relative determination). So far, the coherence between the three methods is satisfactory; the response characteristics for the three determinations are comparable and the agreement between the air density determinations is within 1 × 10⁻⁵ kg m⁻³. A discrepancy of the order of 10⁻⁴ kg m⁻³ is observed between the CIPM-81/91 formula and artefact methods. This difference needs to be clarified by future measurements. The relative combined standard uncertainty obtained on the air density determination when using the air buoyancy artefacts is about 7 × 10⁻⁶.