Bream (Abramis brama) undergo ontogenetic diet shift from zooplankton to benthic macroinvertebrates, but the switching size may be highly variable. To unravel under what conditions bream are pelagic versus benthic foragers, we experimentally determined size-dependent foraging capacities on three prey types from the planktivory and benthivory niche; zooplankton, visible and buried macroinvertebrates. From these data we derived predictions of size-dependent diet preferences from estimates of prey value and competitive ability, and tested these predictions on diet data from the field. Planktivorous foraging capacity described a hump-shaped relationship with bream length that peaked for small bream of 67 mm total length. Benthivory capacity increased with increasing bream size, irrespective if benthic prey were visible on the sediment surface or buried in the sediment. From the experimental data and relationships of metabolic demand we calculated minimum resource requirement for maintenance (MRR) for each of the prey categories used in experiments. MRR increased with bream size for both zooplankton and visible chironomids, but decreased with bream size for buried chironomids, suggesting that intermediate sized bream (120–300 mm) may be competitively sandwiched between small and large bream that are more competitive planktivores and benthivores, respectively. Prey value estimates and competitive abilities qualitatively predicted diet shift in a bream population being released from competition. Competitive release did not change the diet of the largest size-class feeding on an optimal diet of benthic invertebrates both before and after competitive release. However, profound diet shifts towards benthic macroinvertebrates were recorded for intermediate size-classes that fed on a suboptimal diet prior to competitive release. Thus, laboratory estimates of size-dependent foraging capacity of bream in planktivorous and benthivorous feeding niches provided useful information on size-specific competitive ability, and successfully predicted diet preference in the field.