The Poisson, Neyman type A, negative binomial, and the Thomas distributions were fitted to the frequency distributions of the number of fecal groups per plot for mule deer (Odocoileus hemianus) counted on a systematic sample of 960, 100 ft2, and 1,687, 43.6 ft2, circular plots on two diverse Colorado habitats sampled over two seasons. The three latter distributions represented fecal-group data equally well. However, there was a consistent appearance of one or two high frequencies with very small probability of occurrence under the three distributions regardless of plot size, sampling intensity, habitat, or season of fecal-group deposition. Fecal-group densities were too low to distinguish between the three distributions, but the negative binomial offered the simplest explanaffon of ie data, and appropriate standard errors and confidence limits may be readily computed. The two following modes of deer behavior (or a combination of the two) could give rise to any of the three distributions: (1) deer wandering at random over the sampled area in groups of various sizes, and (2) single deer wandering at random within homogeneous blocks of land but a different mean rate for each bloek. Since Bennett English, and McCain (1940) first applied the method to deer (Odocoileus spp.), fecal drol?ping counts have been successfully used as a census method on a statewide basis in Michigan ( Michigan Department of Conservation 1966), on a herd basis in California (McCain 1948), on small study areas in western Texas (Wallmo 1958), and in western Washington ( Brown 1961 ) . Its validity as a census method has been experimentally confirmed with small, known, penned populations of white-tailed deer (O. virginianus) on natural habitat in Michigan (Eberhardt and Van Etten 1956, Ryel 1959); with small, approximately known, free-ranging populations of mule deer in Colorado ( Harris 1959 ) and Montana ( White 1960 ); and for black-tailed deer (O. hemionus columbianus) in California (Dasmann and Taber l9SS). Other uses, not experimentally confirmed) are as indices to the relative intensity of use 1 Contribution from Colorado Federal Aid Projects W-105-R and W-114-R. 2Present address: U. S. Forest Service, Intermountain Forest and Range Experiment Station, Boise, Idaho. and the relative distribution of deer within adjacent habitats ( Wallmo 1958 White 1960) disjunct habitats (Brown 1961)) and portions of fairly homogeneous habitats ( Reynolds 1964 ) . Julander ( 1955 ) used sample counts of deer and cattle fecal droppings as an index to their relative distribution within a diverse habitat in Utah. Neff (1968) presented a detailed review of the literature on the technique and concluded that the method is valid in obtaining reliable big game population data. Few authors have attempted to test the fit of appropriate mathematical models to frequency distributions of fecal group counts. However, lmowledge of an appropriate model is essential for reliable confidence-interval estimates of deer use except perhaps, for very large sample sizes. Knowledge of the model may point to the necessity for transfolming data for further regression and analysis of variance interpretations. Also, appropriate models may offer clues on behavioral patterns. Loveless ( 1967:82) sampled about 150 acres of precipitous winter range with 100ft2 circular plots and concluded that mule