1. Valuable insights about relationships between consumers and their resources have emerged from studies that investigated how these interactions vary with spatial scale. We conducted field studies to examine the scale-dependence of patterns of spatial variation in the densities of larval blackflies (Simulium tribulatum) and water velocity, a key variable determining resource supply rates to these passive suspension-feeders. We sought to answer two main questions: (i) How do patterns of spatial variation in water velocity and larval density change with scale? (ii) Does the relationship between larval density and water velocity vary with scale? 2. We deployed artificial substrates on the bed of riffles and pools to assess spatial variation in water velocity and larval density at three spatial scales spanning four orders of magnitude: (i) among microhabitats within single substrates (centimetre scale); (ii) among substrates within individual riffles or pools (metre scale); and (iii) among riffles and pools (tens of metres scale). We used semi-cylindrical substrates that allowed us to estimate water velocities 1 mm above the substrate surface using a validated computational fluid dynamic model and measurements made with a 10-mm-diameter propeller flowmeter. On natural substrates (stones), we quantified fine-scale velocity variation (measured 1 mm above the stone using hot-film anemometry) and larval density at the two smallest scales. We used hierarchical regression models to partition variation in larval density and water velocity among spatial scales, and to measure the strength of relationships between larval density and water velocity at each scale. 3. The distribution of spatial variation across scales was similar for both water velocity and larval density. The proportion of total variation differed markedly among scales, but variation did not increase monotonically with increasing spatial scale. The greatest proportional variation in velocity and density occurred at the smallest (among microhabitats) and largest (among riffles and pools) scales. 4. The similar cross-scale distributions of variance for water velocity and larval density were not accompanied by consistent density–velocity relationships at all scales. At the smallest scale (i.e. among microhabitats on a single substrate), where spatial variation in velocity was large, there was a consistently strong positive relationship between larval density and water velocity. At the among-cylinder scale (within riffles or pools), where flow variation was smaller, the density–velocity relationship was strongly positive in pools but weakly negative in riffles. At the largest scale (among riffles and pools), the density–velocity relationship was positive among pools, and between riffles and pools, but was negative among riffles.