Objectives Unlike high-income countries, the magnitude of COVID-19-related mortality is largely unknown in many low- and middle-income countries. This study aimed to determine the COVID-19-associated excess mortality in an urban setting in Bangladesh using a cemetery-based death registration dataset. Study design Retrospective observational study Methods Data extracted from the death registry books managed by the local municipality. A total of 6,271 deaths (3,790 male and 2,481 female) recorded between January 2015 and December 2021 were analyzed by using the Bayesian structural time series model (BSTS). Results During the pre-COVID-19 period, the average monthly number of deaths was 69, whereas, during the COVID-19 period, this number significantly increased to 92. Overall, according to model-based results, during COVID-19 period, the number of deaths increased on average by 17% (95% CrI: -18%, 57%): males 29% (95 % CrI: -15%, 75%) and 2.9% for females (95% CrI: -61%, 70%). Conclusions This first-of-its-kind study in Bangladesh has revealed the excess mortality due to the COVID-19 pandemic (2020-2021) in an urban community. It appears that cemetery-based death registration could help track various crises (e.g., COVID-19), especially when collecting data on the ground is challenging for resource-limited countries.