To be able to achieve their physics goals, future neutrino-oscillation experiments will need to reconstruct the neutrino energy with very high accuracy. In this work, we analyze how the energy reconstruction may be affected by realistic detection capabilities, such as energy resolutions, efficiencies, and thresholds. This allows us to estimate how well the detector performance needs to be determined a priori in order to avoid a sizable bias in the measurement of the relevant oscillation parameters. We compare the kinematic and calorimetric methods of energy reconstruction in the context of two ν_μ &8594; ν_μ disappearance experiments operating in different energy regimes. For the calorimetric reconstruction method, we find that the detector performance has to be estimated with an Ο(10%) accuracy to avoid a significant bias in the extracted oscillation parameters. On the other hand, in the case of kinematic energy reconstruction, we observe that the results exhibit less sensitivity to an overestimation of the detector capabilities.