A large area of mangrove wetland has been preciously conserved in Zhenhai Bay of Jiangmen City, Guangdong Province. However, scholars and local government have not paid much attention to this area yet. In this paper, multi-source remote-sensing images of four periods (namely Declassified Intelligence Satellite image acquired in 1970, KH-9, SPOT-2 image obtained in 1999, ALOS multi-spectral image in 2007, and the ZY-3 image captured by domestic satellite in 2015) were used. Integration of spatial distribution characteristics of mangrove forest and objected-oriented multi-scale segmentation, visually interpreted approach were combined for accurate classification of mangrove forest. Results demonstrate that the areas of mangrove forest in Zhenhai Bay were 931.4, 615.3, 678.5 and 783.7 hm2 in 1970, 1999, 2007, and 2015 respectively, showing a declining trend in the former 10 years and later an upward tendency. Specifically, during the period from 1970 to 1999, the change of mangrove forest in this area was mainly due to the destruction of original forest, when only 5.7% of original forest was preserved. Later, during the period of 1999-2015, the interruption of people and artificial remediation coexisted, but on the whole the degree of the damage decreased. The area of mangrove forest increased 63.2 and 105.2 hm2 during 1999-2007 and 2007-2015, respectively. Hence, the primary driving factors that affect mangrove’s temporal and spatial change were the combined influences of the sprawl of aquaculture ponds and the artificial recovery of plants. Therefore, in this paper, we propose a series of measures for targeted protection and development of the mangrove wetland (e.g. protecting the suitable land, developing the ecotourism and realizing the dynamic monitoring on mangroves), so as to provide decision support for rational exploitation and better protection of mangrove forest resources in the area.