In Kenya landslides and mudslides occur mostly during the rainy season and are accelerated by flooding. Flood is a major recurring and seasonal environmental problem in Kenya. These geo-hazards usually cause huge economic losses through destruction of property and loss of life. The increase in size and frequency of floods and landslides are attributed to human activities in the most vulnerable areas and climate change phenomena such as Indian Ocean Dipole. The purpose of this study was to develop flood and landslide susceptibility maps in Tana River, Kenya. The vulnerability analysis provides a clear basis for developing an early warning framework. GIS-based methodologies were used to map the distribution and extent of floods and landslides which include spatial distribution models such as Digital Surface Model (DSM), Digital Elevation Models (DEM) and Landsat Imagery. GIS/RS-based multi-criteria approach was used to select the flood causal factors and risk assessment. Relative weight of each factor was determined using Analytical Hierarchical Process (AHP). The first step in flood vulnerability analysis was to identify the factors. Then each factor’s satellite images were analyzed in the GIS environment, weighted and overlaid to produce the flood vulnerability maps. The results show that the extremely low and minimal flood risk classes cover 11.1 % and 27.4 % respectively. Although very highly vulnerable areas cover the smallest proportion in the study areas (6%) a sizable proportion of areas have substantial risk (20%). The model predicted that a total of 217,882 hectares of land would be inundated during rainy period. Areas that are under high risk in the study area include Garsen North, Mikindu, Chewani and Kipini west. The extremely highly vulnerable areas are: Garsen Central and Garsen South. All these regions are within high and extremely high hazard zones and are dominated by low elevation and slope percent. These vulnerability maps are critical in mapping the flood and landslide events in order to develop a clear roadmap to an early warning system. The adoption of GIS/RS and other machine learning methodologies have proven to be effective tools for hazard forecasting and planning. To achieve the goal of the early warning system, the study recommends the framework with four nodes that require the cooperation between government departments (in county and national government) and non-governmental organizations. This framework consists of four interlinked components: 1) assessments of flood risks areas and local awareness, 2) Flood risk monitoring and warning approaches, 3) Flood risk dissemination and communication service, and 4) Response strategies. The results can be used to support policy options and catchment strategies geared towards ﬂood risk management. Moreover, the spatial vulnerability estimation will help policy makers, local community, and other partners to understand the needed mitigation and adaptation measures. However, further analysis is needed on the extent to which disaster risk reduction is mainstreamed in policy making schemes (environmental, political, socioeconomic, and cultural) especially with the current system of governance with devolved systems.