PROJECT SUMMARY/ABSTRACT This is a study of mortality risk in the deadliest pandemic in the 20th Century, the Spanish flu of 1918, in two of the U.S. cities with the highest number of fatalities, New York and Philadelphia. It has four main aims. Aim 1 is to examine mortality from the Spanish flu at the neighborhood scale in these cities. Aim 2 is to extend the ecological analysis with multilevel models including individual-level predictors. Aim 3 is to compare the patterns of influenza/pneumonia mortality during the pandemic with mortality in the same period in the previous year, and Aim 4 is to compare patterns between cities. Few studies of pandemics have gone beyond analyzing mortality data at the level of whole cities to consider spatial variation within cities, and there have been none that examined risk of mortality in a pandemic using multilevel models. In this project, drawing on individual death records, attention can be directed to specific neighborhoods and mortality risk of their residents. The study will show how personal characteristics such as age, gender, race, nativity, and social class affect risk of mortality in a multilevel model that includes similar characteristics measured at a neighborhood scale and also incorporates spatial relationships among neighborhoods. These aims will be addressed through analysis of a new data set with several innovative components. 1) Transcription of all death certificates during fall 1917 and fall 1918 in New York and Philadelphia including the person’s name, age, race, gender, address, cause of death, date of death, occupation, and parents’ names and place of birth. 2) Geocoding of addresses in a historically accurate GIS system, so that persons can be linked to their neighborhoods. 3) Use of 100% census microdata from 1920 to create measures of neighborhood characteristics, construct a comparison sample of flu survivors, and glean information about the households of those who died. These data will allow the project to probe the independent effects of various individual risk factors while also examining neighborhood differences and spatial patterns in mortality.