Nest fates, causes of nest failure, and identities of nest predators are difficult to determine for grassland passerines. We developed a miniature video-camera system for use in grasslands and deployed it at 69 nests of 10 passerine species in North Dakota during 1996-97. Abandonment rates were higher at nests <1 day after camera deployment (23%) than <1 day after nests were found (2%, P = 0.001). Most birds returned to nests 30 min after camera deployment, but clay-colored sparrows (Spizella pallida) took longer to return to nests than other species (P = 0.035). Cameras did not appear to increase nest predation rates. We videotaped predation by 11 species at 29 nests: eggs or nestlings were destroyed by mice, ground squirrels, weasel, badgers, canids, deer, cowbirds, and hawks. All eggs or nestlings were removed in <15 min at 14 depredated nests. Contents were removed during >1 day or night (22-116 hr) at 6 nests, 5 of which were depredated by ground squirrels or mice. For nests without cameras, estimated predation rates were lower for ground nests than aboveground nests (P = 0.055), but did not differ between open and covered nests (P = 0.74). Open and covered nests differed, however, when predation risk (estimated by initial-predation rate) was examined separately for day and night using camera-monitored nests; the frequency of initial predations that occurred during the day was higher for open nests than covered nests (P = 0.015). Thus, vulnerability of some nest types may depend on the relative importance of nocturnal and diurnal predators. Predation risk increased with nestling age from 0 to 8 days (P = 0.07). Up to 15% of fates assigned to camera-monitored nests were wrong when based solely on evidence that would have been available from periodic nest visits. There was no evidence of disturbance at nearly half the depredated nests, including all 5 depredated by large mammals. Overlap in types of sign left by different predator species, and variability of sign within species, suggests that evidence at nests is unreliable for identifying predators of grassland passerines.