To assess the utility of indigenous habitat knowledge in studies of habitat diversity in Amazonia. Baniwa indigenous communities in Rio Içana, upper Rio Negro, Brazil. Six campinarana vegetation types, recognized and named by a consensus of Baniwa indigenous informants according to salient indicator species, were studied in 15 widely distributed plots. Floristic composition (using Baniwa plant nomenclature only, after frustrated attempts to obtain botanical collection permits), quantitative measures of forest structure and GPS waypoints of the 4-ha composite plot contours were registered, permitting their location on Landsat satellite images. Non-metric multidimensional scaling (NMDS) ordination was carried out using pc-ord software. The NMDS ordinations of the plot data revealed a clear gradient of floristic composition that was highly correlated with three quantitative measures of forest structure: basal area, canopy height and satellite reflectance. Baniwa-defined forest types are excellent predictors of habitat diversity along the structural gradient comprising distinctive white-sand campinarana vegetation types. Indigenous ecological knowledge, as revealed by satellite imagery and floristic analyses, proves to be a powerful and efficient shortcut to assessing habitat diversity, promoting dialogue between scientific and indigenous worldviews, and promoting joint study and conservation of biodiversity.