In this paper the conceptual framework of individual farmers' adoption decisions of new agricultural technologies is used to identify factors that influence adoption modified fanya juu terraces in semi-arid eastern Kenya. The adoption decision model was specified using farm and farmers' characteristics and technology characteristics though likely to influence farmers' adoption behaviour. To test intensity of adoption a Torbit model was specified and estimated. Results of logit regression analysis reveal that access to markets, education, costs, technology attributes, labour, credit and risk preferences significantly influenced adoption decisions. Results from the estimated intensity of adoption model (tobit) show that variation in the proportion of land with technology was significantly influenced by age and level of education of the household head, access to markets, technology attributes, credit, off-farm income and risks.