Evidence in the literature exists to support the theory that mastitis and intramammary infection (IMI) tend to cluster within herds, within cows, and within quarters, facts which may have overarching ramifications on mastitis management in modern dairy herds. Most previous studies, however, have been carried out on prevalent IMI instead of new IMI (NIMI), although reducing incidence of NIMI is a major step toward controlling mastitis. The Canadian Bovine Mastitis Research Network (Saint-Hyacinthe, QC, Canada) has a large mastitis database derived from a 2-yr data collection on a national cohort of dairy farms, and data from this initiative were used to investigate the effect of clustering on the acquisition of NIMI. Longitudinal milk samplings of clinically normal udders taken over several 6-wk periods as well as samples from cows pre-dry-off and postcalving were used (n = 73,772 quarter milk samples). Multilevel logistic models were used to evaluate the effect of location of IMI in quarters of the bovine udder previous to occurrence of an NIMI with Staphylococcus aureus, coagulase-negative staphylococci, and Corynebacterium spp. Several factors were investigated, including the number and location of quarters infected with the pathogen of interest before occurrence of an NIMI, the number of quarters infected with any other pathogen before occurrence of an NIMI (a measure of susceptibility), somatic cell count of the quarter before occurrence of an NIMI, somatic cell count of the other 3 quarters before occurrence of an NIMI, prevalence of the specific pathogen in the herd, and the average somatic cell count of the herd. The amount of variation occurring at different levels (herd, cow, and quarter) for the various pathogens was also calculated. The presence of an IMI in the ipsilateral quarter was associated with an elevated risk of an NIMI occurring for all pathogens investigated. Risk of an NIMI increased considerably as herd prevalence of the pathogen rose. Substantial clustering was found at all levels, with roughly equal amounts of variation found in all 3 levels for coagulase-negative staphylococci, most variation at the cow-level for Corynebacterium spp., and most variation found at the quarter-level for Staph. aureus. Simulation was used to calculate exact values of intraclass correlation coefficients to estimate clustering within cows and within quarters these exact values were, for the most part, lower than estimates calculated using the latent variable approach, but also increased as pathogen prevalence and number of infections in a cow at the previous sampling increased. These results of these analyses can be used to inform approaches to preventing NIMI in modern dairy operations.