Air pollution has changed from an urban environmental problem to a phenomenon spreading to state, country and even global scales. In response, a variety of regulations, standards, and policies have been enacted world-wide. Policy-making and development of efficient and effective control strategies requires understanding of air quality impacts from different sources, which are usually estimated using source-oriented air quality models and their corresponding uncertainties should be addressed. This thesis evaluates emission uncertainties and their impacts on air quality modeling (Models-3/Community Multiscale Air Quality Model (CMAQ)), with special attention to biomass burning. Impacts of uncertainties in ozone precursors (mainly NOX and VOC) emissions from different sources and regions on ozone formation and emission control efficiencies are evaluated using Monte Carlo methods. Instead of running CMAQ multiple of times, first and higher order ozone sensitivities calculated by Higher-order Decoupled Direct method in Three Dimensions (CMAQ-HDDM-3D) are employed to propagate emission uncertainties. Biomass burning is one of the major sources for PM2.5. Impacts of uncertainties in biomass burning emissions, including total amount, temporal and spatial characteristics, and speciation on air quality modeling are investigated to identify emission shortcomings. They are followed by estimation of seasonal PM2.5 source contributions over the southeastern US focusing on Georgia. Results show that prescribed forest fires are the largest individual biomass burning source. Forest fire emissions under different forest management practices and ensuing air quality impacts are further studied. Forest management practices considered here include different burning seasons, fire-return intervals (FRIs), and controlling emissions during smoldering. Finally, uncertainties in prescribed forest fire emissions are quantified by propagation of uncertainties in burned area, fuel consumption and emission factors, which are required inputs for emissions estimation and quantified by various fire behavior models and methods. In summary, this thesis has provided important insights regarding emission uncertainties and their impacts on air quality modeling.