Social scientists have devoted much effort to measuring and analyzing gender and age productivity differentials. In American economic history, gender productivity differentials have important implications for key issues ranging from the relative efficiency of plantations compared to free farms and the pace of industrialization. We use a new data set to estimate direct physical measures by gender and age of productivity in cotton picking—the peak activity and largest use of labor in cotton production. Based on archival data, we have constructed a sample of 755,005 individual observations of daily cotton picking performed by 7,022 enslaved African-Americans on 140 different plantations over 512 plantation-years during the period 1801-1862. Our specific findings include that (1) in the plantation sector, females and males performed essentially equal shares of the picking work over the ante bellum period; (2) before 1840, adult females picked about 2 percent more per day than adult males; (3) after 1840, the differentials reverse and adult males picked 7-11 percent more per day; (4) productivity in picking, performed on an individual basis, was higher on larger-scale units; and (5) the micro picking data raise severe problems for the “pushing” hypothesis recently advanced in the New History of Capitalism literature.