The sub-models for collisions and coalescence are important components of the spray model in multidimensional computations of Diesel sprays. These models influence the computed drop sizes, which affect the overall characteristics of the spray. Typically, the droplet interaction model is separated into two parts: first, calculating the collision rate between particles, and second, calculating the probability of coalescence once a collision has occurred. While the collision frequency may be estimated from kinetic theory considerations, a criterion has to be specified to determine the outcome of the collisions. The outcome may be bouncing, coalescence. stretching separation, or reflexive separation. The coalescence efficiency is defined as the probability that two drops will permanently merge into one drop, given that a collision between the two drops has occurred. Current approaches to modeling the coalescence efficiency are based on experimental observations of binary water drop collisions under atmospheric conditions. However, in the last decade experimental evidence has become available that suggests that the collision behavior of hydrocarbon drops may differ significantly from that of water drops. More recent experiments suggest that the effects of ambient pressure may also be significant. This paper presents a comparison of the computed outcome of drop collisions in a Diesel spray to that of recently published experimental observations. Possible ways to employ the recent findings in multidimensional spray models are discussed. Results of a model modified to reflect this new information is presented and compared with the original model. Limitations of the new model are discussed. (C) 2002 Elsevier Science Ltd. All rights reserved.