The functional significance of dendritic spines and their morphological sensitivity to a wide spectrum of experimental manipulations and pathological states have led to a number of studies in which counts of dendritic spine number have been carried out. These studies have, for the most part, involved the enumeration of only those spines which protrude from the opaque shafts of Golgi-impregnated dendrites into the clear zones flanking the dendrite. Such counts, limited to only those spines which are visible, underrepresent the true total number of spines borne by the dendrites. The magnitude of underrepresentation correlates positively with dendritic shaft diameter and negatively with spine length. This seriously restricts the usefulness of comparisons of spine density between dendrites, or even between segments of the same dendrite. In the present report, a geometrically based method is presented whereby total dendritic spine numbers can be estimated with reasonable accuracy, taking into account factors such as dendrite diameter and spine length. The technique entails the following principal steps: a determination, for a given length of dendrite over which spines are to be enumerated, of the volume of the flanking zones in which spines are visible and can be counted; a determination of the volume of the entire zone which encircles the dendritic shaft and which contains all spines, both visible and not visible; and a proportional extrapolation from the number of visible spines to obtain an estimate of the true total spine number. Tests of the predictive accuracy of the technique using dendrites of known total spine number suggest that estimates which deviate from true total spine numbers by less than 10% can be achieved.