Forests and trees are essential resources for sustainable provision of goods and services. However, trees have not only been depleted in reserved forests but also on agricultural lands. Several measures have been formulated to improve forest cover in Kenya and one of such strategies is agroforestry. This study aimed to examine agroforestry practices, tree density on farms and determine the relationship between socio-economic characteristics of households and tree density on farms in Kaiti watershed, Makueni County, Kenya. Quadrats were used to determine tree density and tree species diversity on farms while questionnaires were used to record socio-economic characteristics of households. Data files were prepared in the Microsoft Excel and SPSS version 20 software where descriptive and inferential statistics were used. The study found 8 agroforestry practices. The results of One-Way ANOVA for both Shannon Diversity Index and Simpsons’ Index of Diversity showed significant difference in species diversity in Kaiti watershed with p-value of 0.00023 and 0.00012, respectively. The mean of tree density was 104.5 trees per acre where 54% of farms had less than 40 trees per acre while 46% of farms had more than 40 trees per acre. Further, the study found significant relationship between socio-economic characteristics of households and tree density on farms. Household income was the most significant with p-value of 0.000. The study recommends sensitization of farmers about importance of trees and suitable tree species for growing in arid and semi-arid areas. Further, supply of certified seedlings close to farmers and at affordable prices would improve tree species diversity and tree density on farms. There is also need for private land ownership for households to promote sense of ownership of trees