Background: Leprosy is known to be unevenly distributed between and within countries. High risk areas or ‘hotspots’ are potential targets for preventive interventions, but the underlying epidemiologic mechanisms of these areas remain not fully understood. In this study, we identified and characterized leprosy hotspots in Bangladesh, a high endemic country. Methods: Data were used from four high-endemic districts in northwest Bangladesh, including 20,623 registered cases (among ~7 million population) between January 2000 and April 2019. Incidences per union were calculated using geospatial population density estimates. A geospatial Poisson model was used to detect incidence hotspots over three (overlapping) 10-year timeframes: 2000-2009, 2005-2014 and 2010-2019. Ordinal regression models were used to assess whether patient characteristics were significantly different for cases outside hotspots, as compared to cases within weak (i.e. relative risk (RR) of one to two), medium (i.e. RR of two to three), and strong (i.e. RR higher than three) hotspots.Results: New case detection rates dropped from 44/100,000 in 2000 to 10/100,000 in 2019. Statistically significant hotspots were identified during all timeframes, and were often located at areas with high population densities. The RR for leprosy was up to 12 times higher for inhabitants of hotspots than for people living outside hotspots. Within strong hotspots (1,930 cases among less than 1% of the population), significantly more child cases (i.e. below 15 years of age) were detected, indicating recent transmission. Cases in hotspots were not significantly more likely to be detected actively. Conclusions: Leprosy showed a heterogeneous distribution with clear hotspots in northwest Bangladesh throughout a 20-year period of decreasing incidence. Findings confirm that hotspots have higher active transmission and are not solely the result of active case finding strategies.