The use of project dispute resolution satisfaction (DRS) as a project success measurement responds to the increasing concern over the spread of the dispute epidemic within the construction industry in Hong Kong. The DRS of a project is considered as favourable where disputes are resolved within the site level. With the use of 48 project data sets and the technique of multivariate discriminant analysis, a discriminant model function was derived to classify projects into favourable DRS and adverse DRS. The multivariate discriminant analysis model function identified eight discriminating variables, among which the degree of design changes offered the highest separation power. The reliability of the discrminating variables was reinforced by a separate relative importance index study. Through the use of principal component factor analysis, these variables were grouped into three factors generically described as substantive influence, facilitation and indirect influence.