Existing work on the impact of business cycles on health yield conflicting results. Studies using aggregate data tend to find strong evidence of pro-cyclicality, while those examining individual-level data find either no effect or counter-cyclicality. We ask whether migration offers an explanation for this, as the aggregate-data approach places strong assumptions about the accuracy of intercensal population estimates. To examine this, we identify a recession that offers plausibly exogenous shock to economic conditions as well as the potential to construct individual mortality data. Our setting is the recession in the cotton textile-producing regions of Britain brought on by the American Civil War. Drawing on comprehensive and publicly available death records and census records we are able to identify where individuals lived at the onset of the recession, regardless of where they died. Results indicate that the mortality rate in these regions increased during the recession. Turning to aggregate data and employing the standard approach suggests that mortality fell during recessions, which is consistent with much of the existing literature. These conflicting results indicate that migration can meaningfully bias estimates of the recession-mortality relationship when using aggregate data.