From what I remember of Social Statistics, standard deviations are used to determine "normal" - i.e. the very middle of the standard deviation graph is the median and 98% of your cases are expected to be within 2 standard deviations (both left and right) from this median. The remaining 2% are three standard deviations away from the median and they are also called outliers. About 50% of the cases are within 1 standard deviation of the median.
One common use for standard deviations used to be (and maybe still is?) calculating grade curves. If you have ever had to put your class on a grading curve, then the median was usually a B- or so, with 50% of the class at that level (clustered around the median), the remaining 46% 2 standard deviations away (Cs and B+/A-s) and the last 2% in the 3rd standard deviation - As and Fs.