Tropical rainforests often have great tree species and functional diversity, complex age and stand structure, deep active sapwood, and potential factors that reduce transpiration, such as frequent cloud cover and wet leaves. Such factors generate challenges for estimating transpiration from sap flux measurements. In fact, upscaling approaches have rarely been attempted for diverse tropical forests. Thus, the objective of this paper is to build upon past efforts to scale transpiration and optimize strategies for wet tropical forests. Over a five-year period, we instrumented 44 trees with heat dissipation sap flow sensors within a premontane wet tropical rainforest in Costa Rica (5000 mm MAP). Given high diversity, few instrumented trees belonged to the same species, genus, or even family. In a subset of trees, radial profiles across the full range of active xylem were examined. Measurements were scaled to the stand in two plots using sapwood area. In association with extensive micro meteorological measurements, only ~10% of rainfall was transpired from this forested watershed due to persistent low radiation, evaporative demand, and frequent wet canopy conditions. Contrary to temperate trees, we observed minimal flow reductions in deeper sapwood among the largest trees, whose water use amounted to nearly 80% of the total stand water use, even though they contributed less than 15% to total stand basal area. We found that transpiration was suppressed when leaves were wet, even after accounting for lower vapor pressure deficit and reduced solar radiation. Since soil moisture conditions at this site remain continually above deficit levels, flow reversal has been assumed to be minor and thus has not yet been investigated. The driest month on record resulted in higher, not lower, transpiration. Given the high diversity of these forests, herein we propose practical and innovative approaches to group trees by function or degree of canopy exposure. These efforts are critical for accurate scaling of measurements from individual sensors to stands and for improving global land surface models that increasingly partition canopy components to better represent these functionally and structurally diverse ecosystems.