This paper contributes new empirical evidence and nuanced analysis on the gender difference in access to extension services and how this translates to observed differences in technology adoption and agricultural productivity. We employ a cross-sectional instrumental-variable regression method using a regionally-representative dataset of more than 7,500 households and 32,000 plots in four major regions in Ethiopia that was collected during the 2010 main season. Results suggest that female heads of households and plot managers are less likely to get extension services and less likely to access quality services than their male counterparts after controlling for plot, household, and village level characteristics. Receiving advice from development agents (DAs) is strongly and positively related to adoption of improved seed and fertilizer for both females and males, as hypothesized. However, beyond their influence through fertilizer and improved seed use, visits by or advice from DAs are not significant in all productivity models estimated for females and males, which is in contrast to past studies. In some crop-specific productivity models estimated, it is the perceived quality of DA visits and access to radio that appear to be strongly and positively significant in explaining productivity levels for both female and male farmers. Our results highlight the need for productivity models that are stratified by gender and crop.