Predicted rates of fire spread using a mathematical model were consistently greater but in reasonably close agreement with rates observed on test fires in ponderosa pine and Douglas-fir slash. Fuel loading, bulk density, particle density, particle surface-to-volume ratio, heat content, total plant salt content, silica-free salt, fuel moisture, and wind velocity were determined as input variables for the test on plots containing slash with and without needles, and at two loadings and three depths. Fuel discontinuities and inadequate knowledge of fuel moisture contents that limit ignition are discussed as the primary reasons for the deviations. One-half of the total fuel weight loss in the flame front was accounted for by particles less than 1 cm. in diameter. A multiple regression for predicting flame length is provided.