Energy flow within TFTR neutral beamlines is measured with a waterflow calorimetry system capable of simultaneously measuring the energy deposited within four heating beamHnes (three ion sources each), or of measuring the energy deposited in a separate neutral beam test stand. Of the energy extracted from the ion source on the well-instrumented test stand, 99.5 ± 3.5% can be accounted for. When the ion deflection magnet is energized, however, 6.5% of the extracted energy is lost. This loss is attributed to a spray of devious particles onto unmonitored surfaces. A 30% discrepancy is also observed between energy measurements on the internal beamline calorimeter and energy measurements on a calorimeter located in the test stand target chamber. Particle reflection from the fiat piate calorimeter in the target chamber, which the incident beam strikes at a near-grazing angle of 12°, is the primary loss of this energy. A slight improvement in energy accountability is observed as the beam pulse length is increased. This improvement is attributed to systematic error in the sensitivity of the energy measurement to small fluctuations in the supply water temperature. An overall accuracy of 15 % is estimated for the total power injected into TFTR. Contributions to this error are uncertainties in the beam neutralization efficiency, reionization and beam scrape-off in the drift duct, and fluctuations in the temperature of the supply water.