Reaction energetics from ab initio calculations are almost always combined with experimental data to derive enthalpies of formation. Some popular strategies are discussed, and a simple criterion is suggested for estimating the accuracy to be expected from using an arbitrary reaction, We recommend using more than one reaction scheme whenever possible and suggest how the results from different reactions may be combined in a weighted average. The uncertainties of the derived values can be estimated from the averaging procedure and by comparing the results obtained using different basis sets and treatments of electron correlation. As examples, we compute the gas-phase thermochemistry for IO-, IO, IO+, and HOI. As a check, analogous calculations are also done for the corresponding bromine species. Finally, we recommend that appropriate experimental data be exploited whenever they are expected to be more accurate than their theoretical counterparts. While compiling the auxiliary data to support these calculations, we found that recent measurements require that the experimental enthalpies of formation of many molecules be revised. We present values for HOCl, HOBr, Cl2O, and Br2O.