The findings show that the distance between associates' homes and locations in the drug production criminal network varied systematically with network characteristics (centrality measures), but not with demographics or criminal history variables. The distance between the geographic locations of associates' homes and the spheres of the majority of their activities was on average 3.96 miles. This suggests that the criminal network involved in drug production is spatially limited. This finding is consistent with the view that criminal associates tend to live closer than farther away from one another, while preferring a "buffer" zone of not being too close to one another. The study also indicated that the drug production location (location for growing marijuana in this case) for the criminal network might keep the participants, and thus their criminal network, tied to a particular geographic area. This supports the central tenets of routine activities theory and pattern theory. In terms of policy and practice, police agencies should maintain ongoing data on geographic crime patterns and the residence locations of known participants in criminal networks. This facilitates the identification and tracking of criminal networks and the individuals involved in it. This is particularly important for the analysis of marijuana production and distribution networks. The study collected data on 376 individuals involved in an illicit marijuana production network in Vancouver, British Columbia (Canada), between 1997 and 2003. The incident, demographic, geographic, and criminal history data were collected from a police records management system. The spatial data were produced with the use of ArcGIS and CrimeStat software packages. 4 tables, 2 figures, and 51 references