Summary We use OLS and decomposition techniques to investigate gender differences in agricultural productivity in Uganda. Using nationally representative surveys from years 2009–2012, the analysis applies different gender dummies – female head of household, female plot holder, and female plot manager- to investigate how the variable of choice affects the calculation of the gender gap. Our analysis obtains different results depending on the gender variable of choice. The study finds that regardless of the variable of choice, the gender gap in agricultural productivity decreases or disappears when factors of production and crop choice are controlled for. The conditional gender gap is about 10% and significant when using female plot manager as the gender variable, while we find no conditional gender gap when using the other gender variables. The use of time fixed-effects and decomposition contributes evidence that the typically available gender variables are insufficient for identifying how gender and decision-making of different household members play a role in productivity. This finding is problematic for targeting effective interventions to increase agricultural productivity and reduce gender inequalities in agriculture. Finally, we find that the older status of female heads, holders and manager of plots, child dependency ratio, and limited access to adult male labor are factors decreasing productivity in female plots in Uganda.