Floodplain plant–herbivore–hydroperiod interactions have received little attention despite their potential as determinants of floodplain structure and functioning. We used five types of exclosures to differentially exclude small-, medium-, and large-sized mammals from accessing Fremont cottonwood (Populus deltoides Marshall subsp. wizlizenii (Watson) Eckenwalder) seedlings and saplings growing naturally on four landform types at an alluvial reach on each of two rivers, the Green and Yampa, in Colorado and Utah. The two study reaches differed primarily as a result of flow regulation on the Green River, which began in 1962. Landforms were a rarely flooded portion of the alluvial plain, geomorphically active slow- and fast-water channel margin sites on the Yampa reach, and an aggrading side channel on the Green. Small-mammal live-trapping and observational data indicated that, with minor exceptions, the kinds of mammals eating cottonwood within each reach were identical. We monitored condition and fates of individual cottonwood plants from October 1993 through the 1997 growing season. Differences in survival and growth were noted both within and between reaches, and both due to, and independent of, mammalian herbivory. Comparisons of cottonwood growth and survivorship among exclosures and between exclosures and controls indicated that a small mammal, Microtus montanus, reduced seedling and sapling survivorship at the Green River reach, but to a lesser extent (seedlings) or not at all (saplings) on the Yampa reach. In contrast, reductions in sapling height increment attributable to medium- and large-sized herbivores were detected only at the Yampa site. We suggest that these differences are a result of (1) flow regulation allowing Microtus populations to escape the mortality normally accompanying the large, snowmelt-driven spring flood, as well as regulation promoting a herbaceous understory favorable to voles, and (2) greater browsing pressure from overwintering deer and elk at the Yampa reach, unrelated to flow regulation. Within areas used by foraging beaver, the probability of a sapling being cut by beaver was similar on the two reaches. This study suggests that changes in riparian plant–herbivore relationships due to shifts in river hydrology may be a common and important consequence of river regulation.