Safety assessment and rational design of bridge structures requires the uncertainty associated with vehicle loads to be modeled as accurately as possible. This modeling is rendered difficult by the presence of vehicle axle weights that involve different combinations of unimodal and multimodal probability distributions with different dependence structures. In this paper, a transformation invariant approach using copula functions is proposed for the multivariate simulation of dependent axle weights of different vehicle classes. Copula based dependence modeling, which is widely used in the financial risk analysis, is applied to model and simulate three different vehicle cases with different combinations of marginal probability distributions for axle weights. The database of observed vehicle weights is based on the data collected at five locations on national highways in India. The dependence between multimodal distributions of axle weights is accurately considered and simulations are carried out. The simulated axle weights are found to be in very good agreement with the observed data. This type of simulation is useful in carrying out simulation-based reliability analysis of bridges and pavements. © 2006 ASCE.