True-coincidence summing correction is an essential element in k0-based NAA1 and becomes important when samples are counted with a high efficiency detector. This may be the case where large detectors are used or where samples are counted in or in the vicinity of the detector in order to achieve low detection limits in conjunction with low-flux reactors. In some laboratories coincidence correction is accomplished by calculating the coincidence correction factors. Since experimental validation of the calculations will reveal only the most significant errors and is a laborious task due to the high number of radionuclides involved, three laboratories decided to compare their calculated coincidence factors. Each laboratory uses a different software package. A comparative performance analysis was made of COINCALC developed at the INW of the University of Gent (implemented in SOLCOI by DSM Research), the software of the IRI, University of Delft, the Netherlands, and the software of the Ecole Polytechnique, Montreal, Canada. The overall approach, data and algorithms were chosen independently by each institute as the software was being developed and, so, the comparison has yielded a number of interesting conclusions. A follow-up investigation of the discrepancies found will probably allow the performance of each program to be improved.