The effectiveness of the training and visit (T&V) extension method in reversing typical biases of the previous conventional system of extension services in Kenya was investigated using multinomial logistic regression analysis. The study provided empirical evidence supporting reduction in biases against small-scale, young and uneducated farmers and remote areas. On the other hand, the bias against marginal production environments and women continued under the T&V system, though at lower rates than before. Efficacy of the T&V system was also con.finned through promotion of higher adoption of improved technologies. This was mainly attributed to the effective mechanisms and better quality of information delivered through the indirect contact methods of T&V.