The number of fixed collectors required to sample the volume and cation content of throughfall precipitation in the coastal forests of British Columbia was examined by means of data from 134 collectors for throughfall volume and 83 collectors for throughfall chemistry. The relative heterogeneity of throughfall parameters in four forest plant associations was also examined. It was concluded that a sampling network suitable for quantifying throughfall volume will be inadequate for quantifying throughfall chemical parameters because of the far greater spatial variability of the latter. This variability is compounded if the study area includes more than one forest plant association. The fixed—collector method was found to be far from ideal for accurate studies of the role of leaching a nutrient cycling because of the excessive number of collectors required. An analysis of the suitability of Wilm's method of roving collectors and Attiwill's modification thereof as alternatives indicated that Wilm's method is inapplicable to studies of throughfall chemistry, but that Attiwill's modification might permit a worthwhile reduction in the number of collectors required in some studies, especially those concerned with gross, long—term studies of leaching. Attiwill's method shares many of the drawbacks of Wilm's method, however, and will probably be unsuitable for many studies. The variability of the chemical data was greatest when expressed as concentrations. Conversion to kg/ha did reduce standard deviations somewhat, although the reduction was modest on all but the most highly variable study area. Sampling at longer intervals did not result in any reduction in the standard deviations of mean throughfall cation concentrations: the effect of longer collection periods on the data expressed as kg/ha was to increase standard deviations significantly. Weekly or more frequent collections are recommended for detailed studies of throughfall chemistry.