This paper describes the development of a breakdown probability model for a long-term work zone in a suburban area. It was found that the Gompertz distribution best describes the breakdown probability at a work zone. In the absence of work zone breakdown probability models, deterministic capacities may still be used for work zone planning because the work zone data sets needed for the construction of breakdown probability models are scarce. The capacity values and the corresponding breakdown probabilities associated with 10 common definitions of deterministic work zone capacity were determined. This information will assist practitioners in choosing work zone capacity definitions that correspond to an acceptable risk of congestion within their jurisdictions. A data mining method was proposed to automate the process for identifying traffic breakdown, and metrics were proposed to assess goodness of fit of various breakdown probability models.