Upland humid tropical forest soils are often characterized by fluctuating redox dynamics that vary temporally and spatially across the landscape. An increase in the frequency and intensity of rainfall events with climate change is likely to affect soil redox reactions that control the production and emissions of greenhouse gases. We used a 24-day rainfall manipulation experiment to evaluate temporal and spatial trends of surface soil (0-20 cm) redox-active chemical species and greenhouse gas fluxes in the Luquillo Experimental Forest, Puerto Rico. Treatments consisted of a high rainfall simulation (60 mm day -1 ), a fluctuating rainfall regime, and a control. Water addition generated high temporal and spatial variation in soil moisture (0.3-0.6 m 3 m -3), but had no significant effect on soil oxygen (O 2 ) concentrations. Extractable nitrate (NO 3 -) concentrations decreased with daily water additions and reduced iron (Fe(II)) concentrations increased towards the end of the experiment. Overall, redox indicators displayed a weak, non-deterministic, nonlinear relationship with soil moisture. High concentrations of Fe(II) and manganese (Mn) were present even where moisture was relatively low, and net Mn reduction occurred in all plots including controls. Mean CO 2 fluxes were best explained by soil C concentrations and a composite redox indicator, and not water addition. Several plots were CH 4 sources irrespective of water addition, whereas other plots oscillated between weak CH 4 sources and sinks. Fluxes of N 2 O were highest in control plots and were consistently low in water-addition plots. Together, these data suggest (1) a relative decoupling between soil moisture and redox processes at our spatial and temporal scales of measurement, (2) the co-occurrence of aerobic and anaerobic biogeochemical processes in well-drained surface soils, and (3) an absence of threshold effects from sustained precipitation on redox reactions over the scale of weeks. Our data suggest a need to re-evaluate representations of moisture in biogeochemical models.