Theory predicts that parents should invest less in dependent offspring with lower reproductive value, such as those with a high risk of predation. Moreover, high predation risk can favor reduced parental activity when such activity attracts nest predators. Yet, the ability of parents to assess ambient nest-predation risk and respond adaptively remains unclear, especially where nest-predator assemblages are diverse and potentially difficult to assess. We tested whether variation in parental investment by a multi-brooded songbird (Brewer's Sparrow, Spizella breweri) in an environment (sagebrush steppe) with diverse predators was predicted by ambient nest-predation risk or direct experience with nest predation. Variation among eight sites in ambient nest-predation risk, assayed by daily probabilities of nest predation, was largely uncorrelated across four years. In this system risk may therefore be unpredictable, and aspects of parental investment (clutch size, egg mass, incubation rhythms, nestling-feeding rates) were not related to ambient risk. Moreover, investment at first nests that were successful did not differ from that at nests that were depredated, suggesting parents could not assess and respond to territory-level nest-predation risk. However, parents whose nests were depredated reduced clutch sizes and activity at nests attempted later in the season by increasing the length of incubation shifts (on-bouts) and recesses (off-bouts) and decreasing trips to feed nestlings. In this unpredictable environment parent birds may therefore lack sufficient cues of ambient risk on which to base their investment decisions and instead rely on direct experience with nest predation to inform at least some of their decisions.