Since the beginning of 1970’s, more than 40 improved maize varieties have been released and disseminated to maize potential areas in Ethiopia. Using cross-sectional survey data collected in 2011 from 39 districts in five Regional States, this paper examines smallholder farmers’ knowledge, adoption and intensity of use of improved maize varieties in the country. Poisson, binary and multinomial Probit, Tobit and Heckman’s selection models are used in explaining determinants of maize variety knowledge, adoption, intensity of maize area under improved varieties at a household level, and type of maize seed used at plot level. Results show that household characteristics, availability of family labor, wealth status, social networks, and access to credit to buy seed and fertilizer, better soil fertility and depth, market opportunities (number of traders known in villages) affect the number of improved maize varieties known to farmers, their adoption and intensity of farm area allocated to improved varieties, and the use of freshly purchased hybrid and/or OPV maize varieties. Generally, institutional arrangements that strengthen farmers’ access to input and output markets and accumulation of wealth could enhance the knowledge and use of improved maize technologies for better productivity and household income.