Stream fish influence macroinvertebrate prey both directly (via consumption) and indirectly through non-lethal effects. Non-lethal effects include life-history responses, such as faster development rates, and behavioral modifications that reduce exposure to predation. Fish predation commonly alters macroinvertebrate communities by reducing vulnerable prey while favoring taxa with behavioral or life-history traits that confer resistance to predation. Such non-consumptive effects of predators have been widely studied in perennial streams but are largely unexplored in non-perennial streams. We examined whole macroinvertebrate assemblages in two small, physically similar non-perennial streams in the Ozark Highlands ecoregion (central USA), one fishless and the other periodically colonized by fish during floods. We asked if any observed differences in community structure were predictable according to our understanding from perennial systems. We quantitatively sampled benthic macroinvertebrates from both streams at two points (Feb and April) during a ~6-month flow season, in two years (2021, 2025). We also collected from two substrate types: exposed bedrock and loose gravel. Between-stream differences strongly outweighed temporal differences within streams and revealed consistently greater diversity in the fishless stream but greater densities in the fish stream. Several mayfly taxa, including ameletids and baetids, contributed to the greater densities in the fish stream, which also showed a greater relative abundance of taxa with traits conferring resistance to predation, such as increased mobility and fast seasonal development. In contrast, non-insects (e.g. amphipods and isopods) and longer-lived insects were more common in the fishless stream. Macroinvertebrate densities were lower on bedrock than loose gravel substrate, and this pattern was amplified in the fish stream. Overall, results suggest that macroinvertebrates in these non-perennial streams respond to fish similarly to expectations from perennial streams, although some of the variation between streams could also be attributed to minor abiotic differences. Future work to continue addressing this question could increase the sample size of fish and fishless streams and include experimental manipulations of fish presence and abundance.