The measured H(D)OCO survival fractions of the photoelectron-photofragment coincidence experiments by the Continetti group are qualitatively reproduced by tunneling calculations to H(D) + CO2 on several recent ab initio potential energy surfaces for the HOCO system. the tunneling calculations involve effective one-dimensional barriers based on steepest descent paths computed on each potential energy surface. the resulting tunneling probabilities are converted into H(D)OCO survival fractions using a model developed by the Continetti group in which every oscillation of the H(D)-OCO stretch provides an opportunity to tunnel. Four different potential energy surfaces are examined with the best qualitative agreement with experiment occurring for the PIP-NN surface based on UCCSD(T)-F12a/AVTZ electronic structure calculations and also a partial surface constructed for this study based on CASPT2/AVDZ electronic structure calculations. These two surfaces differ in barrier height by 1.6 kcal/mol but when matched at the saddle point have an almost identical shape along their reaction paths. the PIP surface is a less accurate fit to a smaller ab initio data set than that used for PIP-NN and its computed survival fractions are somewhat inferior to PIP-NN. the LTSH potential energy surface is the oldest surface examined and is qualitatively incompatible with experiment. This surface also has a small discontinuity that is easily repaired. on each surface, four different approximate tunneling methods are compared but only the small curvature tunneling method and the improved semiclassical transition state method produce useful results on all four surfaces. the results of these two methods are generally comparable and in qualitative agreement with experiment on the PIP-NN and CASPT2 surfaces. the original semiclassical transition state theory method produces qualitatively incorrect tunneling probabilities on all surfaces except the PIP. the Eckart tunneling method uses the least amount of information about the reaction path and produces too high a tunneling probability on PIP-NN surface, leading to survival fractions that peak at half their measured values.