Estimates of the construction time are of key importance in the early phases of the project - they serve as a basis for the decision whether to commence with planning or not, and are used as input for budgets and programmes. Usually, such estimates base on experience with similar projects completed in the past. This experience may be recorded in the form of mathematical models that relate project characteristics to construction time. The aim of the research was, basing on real-life cases, to develop a model of public road building projects duration. The research comprised collection of input for the analyses, preselection of project features correlated with duration, and construction of three models: a simple statistical regression, a multifactor regression and a regression tree. The models were then compared to each other and to the models presented in the literature with respect to their predictive ability. With the assumed set of potential predictors of construction duration, the regression models were found statistically correct, though not precise enough be used as decision-support tools.