Despite extensive vector control programs, dengue remains a significant global health challenge, with outbreaks rising worldwide. Effective dengue control requires reinforcing the surveillance systems and using surveillance data to gain a better understanding of dengue dynamics at both spatial and temporal scales. We studied the effect of socioeconomic and land cover on the presence of dengue hotspots in Medellin (Colombia) from 2010 to 2020 and identified recurrent hotspots during severe epidemic (SE), epidemic (E), and non-epidemic (NE) years. We focused on spatial autocorrelation using global and local indicators of spatial association over 40,814 georeferenced dengue cases. Later, we tested if the spatial units identified as hotspots, recurrent hotspots, and non-hotspots were evenly distributed among socioeconomic strata and land cover categories. During the study period, 50% of the dengue cases were concentrated in 26% of the study area. We identified statistically significant hotspots, some recurring for up to 7 years with their spatial patterns differing between SE, E, and NE years, even though some recurred over time. Recurrent hotspots predominantly occurred in low–medium socioeconomic strata and were absent in the highest strata. Also, they predominated in human-made structures. The interaction between socioeconomic factors, land cover, and potentially, the vector presence seems to explain the spatial variation of dengue epidemics and their recurrent hotspots in Medellin.