OBJECTIVE Body mass index (BMI) may cluster in space among adults and be spatially dependent. Whether and how BMI clusters evolve over time in a population is currently unknown. We aimed to determine the spatial dependence of BMI and its 5-year evolution in a Swiss general adult urban population, taking into account the neighborhood- and individual-level characteristics. DESIGN Cohort study SETTING Swiss general urban population PARTICIPANTS 6,481 geo-referenced individuals from the CoLaus cohort at baseline (age range 35–74 years, period=2003-2006) and 4,460 at follow-up (period=2009-2012). OUTCOME MEASURES Body weight and height were measured by trained health care professionals with participants standing without shoes in light indoor clothing. BMI was calculated as weight (kg) divided by height squared (m2). Participants were geocoded using their postal address (geographic coordinates of the place of residence). Getis-Ord Gi statistic was used to measure the spatial dependence of BMI values at baseline and its evolution at follow-up. RESULTS BMI was not randomly distributed across the city. At baseline and at follow-up, significant clusters of high versus low BMIs were identified and remained stable during the two periods. These clusters were meaningfully attenuated after adjustment for neighborhood-level income but not individual-level characteristics. Similar results were observed among participants who developed obesity. CONCLUSIONS To our knowledge, this is the first study to report longitudinal changes in BMI clusters in adults from a general population. Spatial clusters of high BMI persisted over a 5-year period and were mainly influenced by neighborhood-level income. ARTICLE SUMMARY STRENGTHS AND LIMITATIONS OF THE STUDY • As far as we know, this is the first study to report the persistence of spatial clusters of high BMI values over a 5-year period in adults from a general population • The observed east to west pattern of BMI clustering fits known socio-economic and ethno-cultural differences distinguishing these opposite regions of the city of Lausanne, Switzerland • A consequence of the social policy applied by the city is likely to fix populations with modest income in subsidized housing located in specific areas • While recruitment methods of the CoLaus study aimed at collecting information on a representative sample of the general population, adult participants and non-participants to the CoLaus study may differ and participation bias cannot be excluded • We considered several individual-level covariates but data on individual income was missing. We used instead the median income of the including city statistical sector.