The event-by-event reconstruction procedure and related uncertainties involved in the derivation of excitation energy and source-size distributions are investigated for GeV hadron-induced reactions. The analysis is performed for the 5.0-14.6 GeV/c proton-, π- and antiproton-induced reactions on 197Au, measured with the Indiana silicon sphere charged-particle detector array at the Brookhaven AGS accelerator. The relative contributions of the three major components of the excitation-energy calorimetry: charged-particle kinetic-energy sums, neutrons, and Q values from reconstructed events, are found to be relatively constant for excitation energies above about 500 MeV. Effects on the results imposed by various assumptions necessary to account for experimental factors are examined and a corresponding deconvolution of the excitation-energy distribution is performed. The major uncertainties in the calorimetry are found to be (1) separation of nonequilibrium and thermal-like charged particles, and (2) the unmeasured neutron component. The self-consistency of the procedure is tested via comparisons with the SMM and SIMON codes for the disintegration of hot nuclei.