We extended our previous analyses of term low birth weight (LBW) and preterm birth to 1994-2000, a period of declining air pollution levels in the South Coast Air Basin. We speculated that the effects we observed previously for carbon monoxide, particulate matter < 10 microm in aerodynamic diameter (PM10), and traffic density were attributable to toxins sorbed to primary exhaust particles. Focusing on CO, PM10, and particulate matter < 2.5 microm in aerodynamic diameter (PM2.5), we examined whether varying residential distances from monitoring stations affected risk estimates, because effect attenuation may result from local pollutant heterogeneity inadequately captured by ambient stations. We geocoded home locations, calculated the distance to the nearest air monitors, estimated exposure levels by pregnancy period, and performed logistic regression analyses for subjects living within 1-4 mi of a station. For women residing within a 1-mi distance, we observed a 27% increase in risk for high (> or = 75th percentile) first-trimester CO exposures and preterm birth and a 36% increase for high third-trimester pregnancy CO exposures and term LBW. For particles, we observed similar size effects during early and late pregnancy for both term LBW and preterm birth. In contrast, smaller or no effects were observed beyond a 1-mi distance of a residence from a station. Associations between CO and PM10 averaged over the whole pregnancy and term LBW were generally smaller than effects for early and late pregnancy. These new results for 1994-2000 generally confirm our previous observations for the period 1989-1993, again linking CO and particle exposures to term LBW and preterm birth. In addition, they confirm our suspicions about having to address local heterogeneity for these pollutants in Los Angeles.