Forest cover changes have diverse outcomes for the livelihoods of rural people across the developing world. However, these outcomes are poorly characterized across varying landscapes. This study examined forest cover changes, associated drivers, and impacts on ecosystem services supporting livelihoods in three distinct areas (i.e. remote, intermediate and on-road) in the Chittagong Hill Tracts region of Bangladesh. The three zones had features of decreasing distance to major roads, decreasing levels of forest cover, and increasing levels of agricultural change. Data was collected from satellite images for 1989–2014, structured household interviews, and group discussions using Participatory Rural Appraisal approaches with local communities to integrate and contrast local people's perceptions of forest cover and ecosystem service change with commonly used methods for mapping forest dynamics. Satellite image analysis showed a net gain of forest areas from 1989 to 2003 followed by a net loss from 2003 to 2014. The gain was slightly higher in intermediate (1.68%) and on-road (1.33%) zones than in the remote (0.5%) zone. By contrast, almost 90% of households perceived severe forest loss and 75% of respondents observed concomitant declines in the availability of fuel wood, construction materials, wild foods, and fresh water. People also reported traveling further from the household to harvest forest products. The main drivers of forest loss identified included increased harvesting of timber and fuel wood over time in the intermediate and on-road zones, whereas swidden farming persisted as the major driver of change over time in the remote zone. The contrast between remotely-sensed forest gains and household-perceived forest loss shows community experiences may be a critical addition to satellite imagery analysis by revealing the livelihood outcomes linked to patterns of forest loss and gain. Community experiences may also evoke solutions by characterizing local drivers of forest change. Failing to disaggregate the impacts of forest loss and gains on ecosystems services over time may lead to uninformed management and further negative consequences for human well-being.