This study investigates whether the historical inverse relationship (IR) between land (farm and plot) size and productivity holds for Ethiopia farms. The study uses plot level and household-level data from the three waves of the Ethiopia Socioeconomic Survey. The main finding, which confirms previous studies, is that the plot-size IR holds when productivity measurement is based on self-reported yields. However, the effects were reversed when we used crop-cut yields. Including labor inputs significantly reduces the magnitude of the coefficients on land size, but not the sign. Finally, the quantile regression reveals interesting findings. These are: (1) a strong positive effect of farm (and plot) size on productivity; (2) the magnitude of the effect decreases monotonically with quantile; (3) farm size displays a robust negative impact on gross revenue and the magnitude of the effect increases (in absolute terms) monotonically with quantiles; (4) the effect of farm (and plot) size on productivity decreases in magnitude when we control for labor input; (5) the IR between farm (and plot) size and total and family labor was negative and significant and the effect increases (in absolute terms) monotonically with quantiles.