Livestock production plays a vital role in the global economy, yet little is known about how climate change will affect the industry. Studies of crop agriculture may not translate to livestock agriculture due to differences in the set of potential adaptation strategies. Ranchers' responses to changing weather and climate are particularly relevant in the Brazilian Amazon, where the dry season is increasing by as much as 0.6 days per year. A longer dry season increases production risks as animals may succumb to starvation, lack of water, or heat stress. I pair transaction data and weather shocks to show that ranchers strategically sold animals prior to the dry season, in response to two predictors of an extreme dry season: sporadic rainfall and high temperatures prior to the dry season. Ranchers sold animals for both fattening and slaughter in response to sporadic rainfall prior. Nearly half of animals initially sold for fattening were sold for slaughter 90 days later. In contrast, ranchers responded to high temperatures solely by selling for slaughter, suggesting that current on-farm technologies are insufficient to prevent losses from heat stress. The overall supply increased in years with sporadic rainfall and high temperatures, but the supply fell by a greater amount the following year. My results suggest that farmers are currently not fully insulated from the effects of a severe dry season. As severe dry seasons become more frequent, the industry will likely increasingly rely on feedlots and need additional methods to cope with heat stress.