Access density is defined as the number of accesses per unit length along an arterial. Numerous studies conducted in various regions have indicated that access density has a significant influence on crash occurrences and severities. However, these research findings tend to simplify the relationship between access density and crash attributes and overlook the distinctive local roadway geometric and traffic flow characteristics. This study was conducted to quantitatively understand the impacts of various access densities on the safety performance of major arterials in New Mexico. A cluster analysis and a negative binomial model have been used through computational analysis to investigate the relationship between access density and crash rate. The analysis results demonstrate the piecewise relationship and verify that access density imposes heterogeneous influences on crash rates given different access density ranges, and lower public and commercial access rates are associated with lower crash rates. The impacts of other access features, such as access usage type and median opening type, on crash rates are also investigated. The research findings are helpful to improve safety performance on major arterials in urban metropolitan areas.