Variation in soil properties can influence plant composition and distribution in river floodplains of semi-arid regions. Soil data were collected alongside tree species data on and off the Tana River floodplain in South-eastern Kenya from plots that were organized along transects. A t-test was used to compare soil properties and canonical correspondence analysis to relate the soil properties to the tree species distribution. Moisture (t = 5.92), pH (t = 2.03), P (t = 5.91), Mg (t = 3.25) and Ca (t = 2.04) were significantly higher inside the floodplain (P < 0.05), and bulk density outside the floodplain (t = -8.76, P < 0.05). The influence of soil properties on tree species distribution was higher inside the floodplain with a higher fitted variation of 43% for the first two axes. Bulk density, Ca, Mg, K and CEC significantly influenced the distribution of tree species inside and outside the floodplain. The influence of moisture, P, N and C outside, and pH inside was also significant. The results show that soil properties and flood regime disturbances are important drivers of the vegetation composition. This information is useful in management of tree populations in the light of climate variability and mega dam constructions.