The exposure to the toxic products of the incomplete combustion of wood, charcoal, crop residues and dung used as cooking and heating fuels kills 1.6 million people every year. This leading environmental health risk also accounts for about one-third of the global human-caused black carbon emissions. Stove technologies that vent smoke to the outside of a house and have verified improved combustion efficiencies have been identified as a solution to the household air pollution problem. However, a persistent barrier to success has been the lack of systematic and proven ways to ensure that households continue using the stoves in the long-term and reduce their open fire practices.Over thirty-five years of experience with cookstove dissemination have demonstrated that providing access to the stoves is necessary but not sufficient: people need to accept bringing the stoves into the home (initial acceptance), use them on a the long-term basis (sustained use), incorporate them into their cooking practices, reduce their open-fire use, and importantly, the devices must maintain their performance through time. These seemingly obvious requirements imply that objective metrics need to be developed for these system parameters and ways found to optimize them. The optimization requires better understanding of the cooking technology-behavior interface and an integrative systems perspective to analyze the socio-cultural, economic, ecological and infrastructural factors that regulate the performance of cookstove programs.This dissertation presents a framework of analysis to characterize stove adoption and it introduces the field methods, signal analysis, metrics and visualization tools to quantify the adoption process using low-cost temperature dataloggers as Stove Use Monitors (SUMs) . The SUMs enable the parameterization of stove usage behavior as a critical stove performance parameter that can be objectively measured, unobtrusively monitored and systematically evaluated together with the reductions in air pollution exposures, fuel consumption and greenhouse gas emissions. Data from two biomass chimney-stove case studies are presented: the Guatemala CRECER-RESPIRE stove trial and the Mexico Patsari Stove Project. In the first two chapters I review the main approaches to study the stove adoption process. I propose a new framework that redefines the emphasis previously placed solely on the stove technology, extending it to include the whole cooking system, thereby including the dynamic complex interactions between stoves, fuels and household behavior with the greater socioeconomic and ecological contexts. I argue that at the household level the innovation being introduced is not only the stove technology, but a set of modified or new cooking practices. When a new stove is brought into the home, the interactions between the user and the previous cooking fuels and devices are redefined. Each fuel-device combination creates its own niche and is used for the cooking tasks where it best fits the needs of the user (the adoption niche"). This adoption process commonly leads to the combined use (or "stacking") of new and traditional stoves and fuels and to the redistribution of the cooking tasks that are performed with each device. Drawing from field data and well-documented experiences at the two study sites I find that at the population level the adoption process has three stages: initial adoption, sustained use, and disadoption. Further, I find that the process can be characterized by the following critical parameters: level of acceptance, level of initial use, level of sustained use, time for stabilization, and magnitude of the seasonal fluctuations.In the third chapter I introduce the concept of Stove Use Monitors (SUMs), devices that objectively quantify stove use through direct measurements of physical or chemical parameters on stoves, cookware or food. Using Thermochron iButton temperature dataloggers as SUMs I recorded the initial adoption and sustained use by 82 households in the Guatemala CRECER stove trial. During the 32 months of the study, I recorded stove temperature signals from a total of 31,112 stove-days, with a 10% data loss rate. I present the protocols for sensor placement, data collection and data management specific to these sensors. I implemented a peak detection algorithm based on the instantaneous derivatives and the statistical long-term behavior of the stove and the ambient temperature signals to count the number of daily meals and determine daily usage. The fraction of days in use from all stoves and days monitored in a period (the "percent stove-days") and other key quantitative metrics are defined and their population-averaged behavior and statistical variability are modeled. Using robust Poisson regression models I detected small (3-12%) but statistically significant seasonal variations in the population level of use, while the age of the stove and household size at baseline did not produce statistically significant variations. Usage was highest during the warm-dry period from February to April, with 92% stove-days (95% CI: 87%, 97%) and 2.56 daily meals (95% CI: 2.40, 2.74)