A small population of bald eagles (Haliaeetus leucocephalus) nests along the Rio Yaqui and its tributaries in central Sonora, Mexico (Brown et al., 1987; Brown, 1988). This is the southernmost nesting population of bald eagles known in western North America and the only known inland nesting population of bald eagles in Mexico. Prey use by inland nesting bald eagles in the temperate zone is well known (Lincer et al., 1979), but little information exists on prey of eagles in a subtropical environment. We document prey found in nests of subtropical bald eagles and compare this with prey used in other areas, particularly the endangered population nesting in Arizona. The Rio Yaqui is the largest drainage west of the Continental Divide in Mexico. The Rio Aros and Rio Bavispe, with headwaters in the Sierra Madre of Chihuahua, join to form the Rio Yaqui which flows into the Gulf of California. Three large reservoirs have been constructed in the drainage, two on the Rio Yaqui and one on the Rio Bavispe. The nests of all five known pairs of bald eagles in Sonora are located in or immediately adjacent to riparian habitat within approximately 75 km of the largest impoundment on the Rio Yaqui, the El Novillo Reservoir. Two nests are located on 60-m cliffs; one is in a large hecho cactus (Pachycereus pectin-aboriginum), one is in a fig tree (Ficus petiolaris) growing from a 50-m cliff, and one nest is in a large honey mesquite (Prosopis glandulosa). Subtropical Sinaloan thornscrub (Brown et al., 1980) is the upland vegetation association at all five nest sites. Seven samples of prey remains were collected from three of the five known nests of bald eagles in Sonora. Six samples were taken from nest bowls, and one was taken from the ground beneath a nest during or immediately after the nesting seasons of 1987 to 1989. Remains were identified by comparing diagnostic bones, feathers, or hair of prey with museum specimens (Cash et al., 1985; Haywood and Ohmart, 1986). The minimum umber of prey individuals of each taxon present as equal to the largest number of identical bones per taxon (Mollhagen et al., 1972; Dunstan and Harper, 1975; Haywood and Ohmart, 1986), a tech ique that may underestimate the number of individuals of a taxon taken as prey. The biomass represented by each prey taxon was estimated by multiplying the number of individuals of that taxon by its average body weight. Average body weights of birds and mammals were obtained from Burt and Grossenheider (1964), Dunning (1984), and Haywood and Ohmart (1986). If Dunning (1984) gave the average weight for both sexes of a species, we used the average of the two weights. Weight estimates for large doves of unknown species were the average of mean weights of white-winged dove (Zenaida asiatica) and white-tipped dove (Leptotila verreauxi), the two largest doves in the study area. Regression equations for body length and weight of fish prey were obtained by measuring either the opercle, pectoral spine, or skull length of fresh or preserved specimens of known length and weight (Hansel et al., 1988). Using either opercle length, pectoral spine length, or skull length of the prey remains, regression equations were used to calculate standard length and weight of prey. Standard length is the distance from the most anterior part of the fish to the posterior end of the vertebral column. Samples of common carp (Cyprinus carpio, an introduced species), river carpsucker (Carpiodes carpio, an introduced species), and catfish (Ictalurus sp., one introduced and one native species) taken from the Rio Yaqui were measured, preserved, and used in addition to other, preexisting museum specimens for length and weight regressions. Measurements from preserved specimens of Sonora sucker (Catostomus insignis) were substituted in the regression analysis for Yaqui sucker