A fundamental issue in speech research concerns whether distinctions in terms of place of articulation are more succesfully captured by local (static) or global (and/or dynamic) properties of the speech signal. While most studies of place of articulation have investigated stop consonants, it is uncertain whether these properties can successfully classify fricatives. In the present study, both static and dynamic metrics were used to investigate place of articulation in fricatives. Static metrics include spectral peak location, noise duration, and noise amplitude; dynamic metrics include relative amplitude, locus equations, and spectral moments. Twenty speakers produced the fricatives /f, v, θ, ð, s, z, ∫, ȝ/ followed by the vowels /i, e, ae, ɑ, o, u/. Preliminary results based on ANOVA and discriminant analysis suggest that all metrics could distinguish nonsibilant /f, v, θ, ð/ from sibilant /s, z, ∫, ȝ/. In addition, spectral moments separated /s, z/ from /∫, ȝ/ with about 95% accuracy and /f, v/ from /θ, ð/ with about 65% accuracy while locus equations only served to separate labiodental /f, v/ from all other fricatives. Accuracy of the metrics will also be compared to human perception data. [Work supported by NIH.]