AbstractThe International Roughness Index (IRI) is widely used in evaluating pavement condition, making repair decisions, assessing ride comfort, and estimating vehicle operating costs. However, it is generally costly to measure IRI, and for this reason, certain road classes are excluded from IRI measurements at the network level. It is hypothesized that it is feasible to estimate the IRI of a pavement section given its distress types and their respective densities and severities. To investigate this hypothesis, this paper uses data from in-service pavements in a Midwestern US state and multiple statistical and machine learning techniques, namely least absolute shrinkage and selection operator (Lasso) and Ridge regression, support vector regression (SVR), regression tree, and random forests methods. These techniques were used to ascertain the extent to which IRI can be predicted given a set of pavement attributes. The data set contains comprehensive disaggregate data on pavement performance (IRI) and distress variables (rutting, faulting, texture, and cracking) collected by automated equipment. The analysis results suggest that it is feasible to estimate reliable IRI at a pavement section based on the distress types, densities, and severities at that section. The results also suggest that such estimated IRI is influenced by the pavement type and functional class. The paper also includes an exploratory section that uses Gaussian techniques to address the reverse situation, that is, estimating the distribution of extant pavement distress types, severity, and extent based on the roughness value of that section.