Context. The census of the globular clusters (GCs) in the Milky Way (MW) is still a work in progress. The advent of new deep surveys has made it possible to discover many new star clusters both in the Galactic disk and bulge, but many of these new candidates have not yet been studied in detail, leaving a veil on their true physical nature. Aims. We explore the nature of 19 new GC candidates in the Galactic bulge by analysing their colour-magnitude diagrams (CMDs) in the near-infrared (NIR) using the VISTA Variables in the Via Láctea Survey (VVV) database. We estimate their main astrophysical parameters: reddening and extinction, distance, total luminosity, mean cluster proper motions (PMs), metallicity, and age. Methods. We obtain the cluster catalogues including the likely cluster members by applying a decontamination procedure on the observed CMDs based on the vector PM diagrams from VIRAC2. We adopt NIR reddening maps in order to calculate the reddening and extinction for each cluster, and then estimate the distance moduli and heliocentric distances. Metallicities and ages are evaluated by fitting theoretical stellar isochrones. We also calculate their luminosities in comparison with known Galactic GCs. Results. We estimate a wide reddening range of 0.25 E(J − K s ) 2.0 mag and extinction 0.11 A K s 0.86 mag for the sample clusters, as expected in the bulge regions. The range of heliocentric distances is 6.8 D 11.4 kpc. This allows us to place these clusters between 0.56 and 3.25 kpc from the Galactic centre, assuming R = 8.2 kpc. Also, their PMs are kinematically similar to the typical motion of the Galactic bulge, apart from VVV-CL160, which shows different PMs. We also derive their metallicities and ages, finding −1.40 [Fe/H] 0.0 dex and t ≈ 8 − 13 Gyr respectively. The luminosities are calculated both in K s − and V−bands, recovering −3.4 M V −7.5. We also examine the possible RR Lyrae members found in the cluster fields. Conclusions. Based on their positions, kinematics, metallicities, and ages, and comparing our results with the literature, we conclude that nine candidates are real GCs, seven need more observations to be fully confirmed as GCs, and three candidates are discarded as GCs and appear to be younger open clusters.