Monitoring of wild populations is central to species conservation and can pose a number of challenges. To identify trends in populations of parrots, monitoring programmes that explicitly take detectability into account are needed. We assessed an occupancy model that explicitly accounted for detectability as a tool for monitoring the large macaws of Bolivia's Beni savannahs: the blue-throated Ara glaucogularis, blue-and-yellow Ara ararauna and red-and-green macaws Ara chloropterus. We also evaluated the joint presence of the three macaw species and estimated their abundance in occupied areas. We modelled occupancy and detection for the three macaw species by combining several site and visit covariates and we described their conditional occupancy. Macaws occupied two thirds of the surveyed area and at least two species occurred together in one third of this area. Probability of detection was 0.48–0.86. For each macaw species, occupancy was affected by the abundance of the other two species, the richness of cavity-nesting species, and the distance to the nearest village. We identified key priority areas for the conservation of these macaws. The flexibility of occupancy methods provides an efficient tool for monitoring macaw occupancy at the landscape level, facilitating prediction of the range of macaw species at a large number of sites, with relatively little effort. This technique could be used in other regions in which the monitoring of threatened parrot populations requires innovative approaches.