In a search for spectral parameters that can be used to distinguish and to model fricatives, spectral moments, dynamic amplitude, and slope above maximum amplitude were computed for a fricative corpus including sustained fricatives at different effort levels, and fricatives in vowel context. Moments varied significantly by frequency range used in computation. M3 appeared to vary the least across fricative, contrasting with Forrest et al.'s 1988 study. Dynamic amplitude separated sibilants and non-sibilants, as predicted; slope above the maximum amplitude varied significantly with effort level.