Worldwide, loneliness is one of the most common psychological phenomena among older adults, adversely affecting their physical and mental health conditions during the COVID-19 pandemic. This study aims to assess changes in the prevalence of loneliness in the two timeframes (first and second waves of COVID-19 in Bangladesh) and identify its correlates in pooled data. This repeated cross-sectional study was conducted on two successive occasions (October 2020 and September 2021), overlapping with the first and second waves of the COVID-19 pandemic in Bangladesh. The survey was conducted remotely through telephone interviews among 2077 (1032 in the 2020-survey and 1045 in the 2021-survey) older Bangladeshi adults aged 60 years and above. Loneliness was measured using the 3-item UCLA Loneliness scale. The binary logistic regression model was used to identify the factors associated with loneliness in pooled data. We found a decline in the loneliness prevalence among the participants in two survey rounds (51.5% in 2021 versus 45.7% in 2020; P = 0.008), corresponding to 33% lower odds in the 2021-survey (AOR 0.67, 95% CI 0.54-0.84). Still, nearly half of the participants were found to be lonely in the latest survey. We also found that, compared to their respective counterparts, the odds of loneliness were significantly higher among the participants without a partner (AOR 1.58, 95% CI 1.20-2.08), with a monthly family income less than 5000 BDT (AOR 2.34, 95% CI 1.58-3.47), who lived alone (AOR 2.17, 95% CI 1.34-3.51), with poor memory or concentration (AOR 1.58, 95% CI 1.23-2.03), and suffering from non-communicable chronic conditions (AOR 1.55, 95% CI 1.23-1.95). Various COVID-19-related characteristics, such as concern about COVID-19 (AOR 1.28, 95% CI 0.94-1.73), overwhelm by COVID-19 (AOR 1.53, 95% CI 1.14-2.06), difficulty earning (AOR 2.00, 95% CI 1.54-2.59), and receiving routine medical care during COVID-19 (AOR 2.08, 95% CI 1.61-2.68), and perception that the participants required additional care during the pandemic (AOR 2.93, 95% CI 2.27-3.79) were also associated with significantly higher odds of loneliness. However, the odds of loneliness were significantly lower among the participants with formal schooling (AOR 0.71, 95% CI 0.57-0.89) and with a family of more than four members (AOR 0.76, 95% CI 0.60-0.96). The current study found a decreased prevalence of loneliness among Bangladeshi older adults during the ongoing pandemic. However, the prevalence is still very high. The findings suggest the need for mental health interventions that may include improving social interactions increasing opportunities for meaningful social connections with family and community members and providing psychosocial support to the vulnerable population including older adults during the pandemic. It also suggests that policymakers and public health practitioners should emphasise providing mental health services at the peripheral level where the majority of older adults reside.