The paper considers, through modelling, how software faults are spread throughout the entire life-cycle of a large software product and how fault detection and correction processes will affect the spreading mechanism. The study is further enlarged to incorporate models for cost estimation. The models can be described as being of a qualitative rather than a quantitative nature, in that they highlight the effects of different approaches relative to each other before giving ‘exact’ values for each approach. The study reveals that the behaviour and consequences of different ways of spreading and detection, as well as different cost mixtures, can be studied and thus understood.