This paper presents the results obtained from detailed studies of the properties of smoke particles produced from a wide range of flaming and non-flaming combustible materials and discusses how these properties impact early-warning fire detection as well as the hazards of smoke particle toxicity and reduced visibility that can significantly affect life safety. Data acquired include discrete angular scattering at wavelengths of 635 nm and 532 nm; visible light obscuration; light extinction at a wavelength of 532 nm and total light scattering at a wavelength of 520 nm; the responses of calibrated combination ionization/photoelectric smoke sensor; and total mass concentrations. These data are subsequently used to define the size, morphology and radiative transfer properties of the fractal aggregate smoke particles including radius of gyration, primary particle diameter, number of primary particles per aggregate, mass of an aggregate, mass extinction, scattering and absorption coefficients and the resultant albedo. Scanning electron microscope (SEM)/transmission electron microscope (TEM) data and computer-generated fractal aggregates are compared to determine similar morphologies and then used to calculate theoretical values of scattering, absorption, and extinction efficiencies using both the discrete dipole approximation (DDA) and the Rayleigh-Debye-Gans (RDG) approximation for subsequent comparison to the experimental data. These data and analyses indicate that significant differences exist between flaming and non-flaming smoke particles in terms of size, morphology and radiative transfer properties. From a practical viewpoint, the analyses also indicate possible techniques for development of improved early warning fire sensors and smarter, discriminating fire sensors that can function in hostile, contaminated atmospheres such as mines and tunnels. These atmospheres may contain significant levels of combustion products from internal combustion engines, such as diesels, that are used routinely in underground mines. In addition, the much higher albedos measured for non-flaming smoke particles are indicative of significantly lower carbon content and higher levels of volatile organic compounds that have the potential for increased acute toxicity due to their higher reactivity. The paper demonstrates how the basic data can be used to implement improved fire detection systems and improve our ability to assess hazards resulting from potentially catastrophic mine fires.