While acoustic-to-articulatory mapping has been successful for purely voiced speech, a complete acoustic-to-articulatory mapping solution for all classes of speech sounds remains to be found. This work investigates extending acoustic-to-articulatory mapping techniques to fricative sounds. A normalized spectral correlation measure has been developed for an acoustic-to-articulatory mapping routine that produces good acoustic fits and reasonable articulatory configurations for isolated fricatives. The mapping routine can, by using this distance measure along with amplitude information, effectively separate fricative classes by place of articulation. Based on results with isolated fricatives, the acoustic-to-articulatory mapping of fricatives has been extended to the intervocalic case, where information in the voiced segments preceding and following a fricative is used to initialize its optimization. Confidence in the acoustic-to-articulatory mapping of fricatives in isolation influences how they are used in this dynamic case. Heuristics for controlling frication amplitude and mixed excitation are discussed.