There are thousands of subterranean coal fires in the world that, because of incomplete combustion, emit a wide variety of volatile and semivolatile organic compounds to the atmosphere, water, and soil at concentrations that could pose health risks to humans and wildlife. The main goals of this study were to (1) review methods that are used to characterize physical and chemical characteristics of coal-fire sites, (2) determine relationships between gas emissions and physical and chemical characteristics of coal-fire sites, using a combination of regression and multivariate statistical methods, and (3) determine the concentrations of volatile and semivolatile organic compounds in water and soil at two coal-fire sites in eastern Kentucky. More specifically: The objective of Chapter 1 was to review past works and list technologies used over time. Eight years of coal-fire collection technologies were reviewed. A variety of methods and technologies were identified. Qualitative and quantitative preferences were noted. The objective of Chapter 2 was to identify and list uncontrolled coal-fire variables. These variables include complete/incomplete combustion; fire temperature and size; distance to fire; relative humidity and moisture in the system; geology, geochemistry, and age of coal; condition of the mine, sampling time of day; sampling equipment differences; and human error. A secondary objective of this chapter was to determine which coal-fire gases have strong relationships by using the principal component analysis (PCA) software JMP. The strongest relationship was between CO and H2S. Temperature and CH4 were also important. This indicates that incomplete combustion and polynuclear aromatic hydrocarbon (PAH) formation are likely occurring, setting the stage for Chapter 3. The objective of Chapter 3 was to identify and define the extent of soil and water hydrocarbon contamination at the Truman Shepherd and Lotts Creek coal fires in eastern Kentucky. No groundwater contamination was detected at either location. Soil contamination was found at both, but was much higher at Lotts Creek, potentially because of sorption onto soil organic matter (which is reduced at Truman Shepherd by an excavation attempt) and other physicochemical mechanisms. Soil contamination was localized to relatively small areas around coal-fire vents. Based on the results, future studies should consider: Attempting to duplicate these results in other geologic regions Quantifying greenhouse gas emissions from coal fires to consider their contribution to climate change. Coal-fired power plants are regulated, but coal fires, which produce many more harmful gases, are not Determining the feasibility of an oxygen-injection system to engender more complete combustion, therefore possibly reducing harmful gases Determining the feasibility of electricity production from coal fires Adopting a consistent federal coal-fire policy