AbstractFarm succession is a central issue in agricultural policy. Yet although many studies explore succession planning, little is known about how farms are actually transferred. We provide the first population‐level evidence on intergenerational farm succession by linking US census records for millions of farmers' children in 1900 and 1910 to identify which children own and operate the family farm up to 40 years later. We first show that daughters are rarely successors. Using a within‐family identification strategy, we find that first‐born sons are slightly more likely than their younger brothers to be successors while their parents are working aged. However, birth order is not predictive of who receives the farm when parents are older or deceased. For later farm transfers, sons who were previously tenant farmers are much more likely than their brothers to be successors, possibly because they are better prepared. Fewer than one‐fifth of farmers transfer their farm to any son. Our study relies on rich historical data because current policy prevents the necessary data linkages for studying intergenerational farm succession. Providing a secure system for researchers to link modern agricultural data to population microdata, similar to what has been achieved with other data sources, would yield crucial insights into long‐term agricultural policy issues in the US.