This work develops guidance and tools to understand the performance, improve the design, and simplify the evaluation of naturally ventilated low-rise commercial buildings in warm and hot climates. We conducted 50,000 detailed energy and airflow simulations in 427 locations across Brazil, varying 55 parameters representing building morphology, fenestration, construction properties, internal gains, operating times, wind modifiers, flowpaths, window control, and soil traits. Comfort performance was quantified by the average annual fraction of occupied hours that exceeded the upper limit of an adaptive comfort zone, and investigated with sensitivity analysis and machine learning methods. Results indicated that, after climate, building size (both footprint area and number of stories) and internal gains were most influential and were positively associated with discomfort. Adding air movement with ceiling fans and providing for night ventilation both proved highly effective comfort interventions. Except for roof solar absorptance, opaque envelope changes, including increasing insulation or thermal mass, had only marginal impacts. A support vector regression metamodel, requiring 29 easily obtainable inputs plus a weather file, was fit to the simulation results and successfully validated (R-2 = 0.97). The metamodel was developed as a simplified compliance path for naturally ventilated buildings to enhance Brazil's commercial building performance labeling program, which, because it currently provides such a path only for air conditioned buildings, may discouraging decision-makers from considering even more efficient passive solutions. We use a case study to show how the metamodel, which we will distribute publicly, can also serve as a design tool, and demonstrate that modifying a small set of parameters can drastically improve thermal performance and achieve sustainable comfort in hot and warm climates. (C) 2016 Elsevier Ltd. All rights reserved.