The probability model of software errors offered as a partial justification for the modified condition/decision coverage (MC/DC) method of testing software is examined. MC/DC is complex and still evolving, but a description of MC/DC adequate for a discussion of the probability model is presented. The original description of the model is presented, and the model's relationship to MC/DC is described. The model is reformulated to make it easier to understand, and it is generalised to strengthen it. The assertions about MC/DC and programming that have been made on the basis of the model are examined. Finally, some experiments are proposed to estimate the model's parameters and to test the appropriateness of the model.