Tropical forests provide critically important ecosystem services, and they are particularly important for their levels of biodiversity and for the carbon that they store. Yet despite global efforts to slow or halt deforestation, natural forests in the tropics continue to be cleared, primarily for agricultural expansion. Indonesia contains the world’s third largest humid tropical forest area, and for much of the past several decades has experienced alarmingly high rates of deforestation. This began to change in 2017, when deforestation rates dropped precipitously and have since remained low. To better understand how this recent trend compares to historical deforestation patterns, this study used a sample-based approach to estimate annual primary forest loss in Indonesia over a 30-year period, from 1991-2020. Since 1990, Indonesia has lost 28.4 (standard error of +/-0.7) Mha of primary forest – roughly one quarter of its total primary forest area in 1990. One fifth of this area (19.7% +/-1) was cleared during a single two-year period, 1997 and 1998, when millions of hectares of primary forest were burned during a severe El Niño event. I also tracked land use after forest clearing to better understand what drives deforestation in Indonesia and found that more than half of all forests were left idle after clearing, often for years at a time. While some of this was caused by forest fires, like those that occurred during the 1997/98 El Niño event, the majority, 8.5 (+/-0.4) Mha, was actively cleared. Large areas of actively and fire-cleared land remained unused at the end of the study period (4 +/-0.3 and 4.8 +/-0.3 Mha, respectively). However, by 2020, an estimated 40.7% (+/-1.7) of initially unproductive land had also been converted to productive land uses, primarily palm oil production, which covered 16 (+/-0.5) Mha of land in Indonesia in 2020. This included 2.5 (+/- 0.2) Mha of land used to cultivate oil palms that directly replaced primary forests and another 5.3 (+/-0.3) Mha that expanded into previously forested areas one or more years after forest conversion. In the last few years of the study, my sample-derived estimates also confirmed a decline in deforestation after 2016, which had previously been seen in forest loss estimates derived from map pixel counting. From 2017-2020 Indonesia experienced the lowest rates of primary forest clearing observed during the study period. This drop in deforestation occurred after years of increasingly tight restrictions related to primary forest conversion, peatland use, and palm oil expansion, and during a period of heightened public concern about deforestation and land fires following the 2015 El Niño event. It also occurred during a time when palm oil prices were relatively low, and after millions of hectares of idle land had been intentionally created, a phenomenon that is likely closely tied to speculation and land banking. This study provides the most detailed information currently available about historic deforestation trends and land use trajectories after forest clearing in Indonesia, shedding new light on forest change patterns and providing a dataset that could potentially be used in future studies, including for econometric research to quantify the extent to which political and economic factors may have influenced land cover change.