Experts in health, recreation, planning, and transportation are collaborating in studies of the effects of urban design on physical activity. One particular area of study is the use of multiuse urban greenway trails for recreation, fitness, transportation, and other utilitarian uses. Huston, Evenson, Bors, and Gizlice (2003, p. 69) conclude, "… there seems to be preliminary support for the assertion that presence of trails may increase physical activity levels…". This finding is important because it provides health-based evidence for features of urban design that heretofore have been justified solely on the bases of recreation or transportation. Additional evidence is needed to confirm the effects of trails on physical activity. This paper presents new empirical results on neighborhood correlates of traffic on multiuse urban greenway trails based on traffic counts taken at 27 locations in Indianapolis, Indiana using infrared sensors. The results include both descriptive information about traffic patterns on trails and models that correlate trail traffic with neighborhood characteristics, including dimensions of the built environment, socio-demographic neighborhood characteristics, and other variables such as weather. The data set includes results from continuous monitoring at four locations on one trail since February, 2001; two locations on a second trail since July, 2002; and 21 locations on five trails since May, 2004. The descriptive results include average daily traffic, average peak hour traffic, and peak hour congestion factors for all 27 monitoring locations. Daily traffic is modeled as a function of both policy relevant and other variables. Policy relevant variables include dimensions of the built environment such as density of street network in contiguous neighborhoods; land use mix (e.g., percent commercial); and the normalized difference vegetation index (NDVI), an indicator of the distribution of vegetation along trails and in contiguous neighborhoods. Socio-demographic variables include population density, socioeconomic status (e.g., median household income and educational attainment), and thnicity. Other explanatory variables are month, day of week, and weather, including precipitation, deviation from long term daily average temperature, and percent sunshine. These models build on a set of log-linear models developed for four locations on one trail that explain up to 85% of the variation in daily trail traffic at the different locations (Han and Lindsey 2004). These results show that traffic at the different locations is significantly different and that that higher median household income, higher population, and greater proportions of white residents are associated with higher traffic counts on trail segments. The addition of additional locations and new trails to the dataset adds degrees of freedom to the models and the ability to account for additional neighborhood influences. Health planners can use these new models to understand how the public responds to the facilities and to make more effective decisions about interventions to increase physical activity.