Based on the Landsat images in 2000 and 2016, the remote sensing ecological index (RSEI) is employed to estimate the eco-environment state of Fuzhou City, then the response mechanism of the RSEI to the road network is quantitatively explored using three different sampling strategies, including road buffer zone, urban-rural gradient zone, and section line. Moreover, the spatial heterogeneity of road kernel density (KDE), RSEI and their relationships are analyzed using globally spatial autocorrelation analysis and geographically weighted regression analysis, respectively, in different spatial scales of sample units, including 500 m×500 m, 1000 m×1000 m, 1500 m×1500 m, 2000 m×2000 m, 2500 m×2500 m, and 3000 m×3000 m. The results show that the areas in fine state of eco-environment are greater than those in poor status, indicating an improvement of the eco-environment quality from 2000 to 2016 in the study area. The RSEIs in various types of road buffers are steadily increasing from 0 m to 3000 m distances to roads. Among which the influence thresholds of the national road, provincial road, county road and township road are about 900 m, 900 m, 450 m and 750 m, respectively. In urban-rural gradient analysis, the curve of RSEI increases with the increasing distances from the administrative center, and it tends to be gentle after reaching a certain threshold, which is around 20 km in the district-level, and 12 km in the county-level, respectively. However, the trend of KDE is exactly opposite of the RSEI's, but with the similar thresholds. Along the section line, the RSEI is relatively low, while the KDE is relatively high in the administrative center; and the RSEI of the inland area in the northwest is higher than that of the coastal area in the southeast. Spatial autocorrelation analysis shows that the RSEI and the KDE have a higher spatial aggregation effect at the two sampling units of 1500 m×1500 m and 2000 m×2000 m than that of the other sampling units, therefore, 1500 m×1500 m and 2000 m×2000 m are then employed as spatial analysis units to explore the spatial variations in the relationships between RSEI and KDE using geographically weighted regression (GWR) model. The GWR outcomes overall indicate a negative correlation between RSEI and KDE, but also identify the spatial paradigms in their divergent correlations, with the negative associations mainly distributing in the central area of the study area. Our study can provide scientific basis for the ecological civilization construction and sustainable development of road network for the study area.