Researchers in the field of turbulent combustion have found fractal geometry to be a useful tool for describing and quantifying the nature of turbulent flames. This paper describes and compares several techniques for the fractal analysis of two dimensional (2-D) turbulent flame images. Four methods of fractal analysis were evaluated: the Area Method, the Box Method, the Caliper Method, and the Area-Caliper Method. These techniques were first applied to a computer-generated fractal image having a known fractal dimension and known cut-offs. It was found that a “window” effect can cause the outer cut-off to be underestimated. The Caliper Method was found to suffer from noise arising from the statistical nature of the analysis. The Area-Caliper Method was found to be superior to the other methods. The techniques were applied to two types of flame images obtained in a spark ignition engine: Mie scattering from particles seeded in the flow and laser induced fluorescence of OH. Both imaging techniques yielded the same fractal dimension for images obtained at the same engine operating conditions. A “fractal thickness” is described. This is the range of threshold intensities defining the flame boundary over which the fractal dimension is constant.