Soil carbon monoxide (CO) fluxes represent a net balance between biological soil CO uptake and abiotic soil and (senescent) plant CO production. Studies largely from temperate and boreal forests indicate that soils serve as a net sink for CO, but uncertainty remains about the role of tropical rainforest soils to date. Here we report the first direct measurements of soil CO fluxes in a tropical rainforest and compare them with estimates of net ecosystem CO fluxes derived from accumulation of CO at night under stable atmospheric conditions. Furthermore, we used laboratory experiments to demonstrate the importance of temperature on net soil CO fluxes. Net soil surface CO fluxes ranged from −0.19 to 3.36 nmol m−2 s−1, averaging ∼1 nmol CO m−2 s−1. Fluxes varied with season and topographic location, with the highest fluxes measured in the dry season in a seasonally inundated valley. Ecosystem CO fluxes estimated from nocturnal canopy air profiles, which showed CO mixing ratios that consistently decreased with height, ranged between 0.3 and 2.0 nmol CO m−2 s−1. A canopy layer budget method, using the nocturnal increase in CO, estimated similar flux magnitudes (1.1 to 2.3 nmol CO m−2 s−1). In the wet season, a greater valley ecosystem CO production was observed in comparison to measured soil valley CO fluxes, suggesting a contribution of the valley stream to overall CO emissions. Laboratory incubations demonstrated a clear increase in CO production with temperature that was also observed in field fluxes, though high correlations between soil temperature and moisture limit our ability to interpret the field relationship. At a common temperature (25 °C), expected plateau and valley senescent-leaf CO production was small (0.012 and 0.002 nmol CO m−2 s−1) in comparison to expected soil material CO emissions (∼ 0.9 nmol CO m−2 s−1). Based on our field and laboratory observations, we expect that tropical rainforest ecosystems are a net source of CO, with thermal-degradation-induced soil emissions likely being the main contributor to ecosystem CO emissions. Extrapolating our first observation-based tropical rainforest soil emission estimate of ∼ 1 nmol m−2 s−1, global tropical rainforest soil emissions of ∼ 16.0 Tg CO yr−1 are estimated. Nevertheless, total ecosystem CO emissions might be higher, since valley streams and inundated areas might represent local CO emission hot spots. To further improve tropical forest ecosystem CO emission estimates, more in situ tropical forest soil and ecosystem CO flux measurements are essential.