In order to tell whether an individual child is overweight or underweight, the simplest method is to plot his or her growth measurements on a growth chart and interpret the changes in percentiles from the chart.
If you'd like to calculate percentiles exactly, you can use computer programs. One is Epi Info 2000, a program that allows you to enter and analyze data. The anthropometric component of Epi Info 2000, Nut-Stat, can be used to calculate percentiles exactly. The program can be downloaded at no cost from www.cdc.gov/epiinfo.
For Statistical Analysis System (SAS) users, CDC has written a code that calculates percentiles (www.cdc.gov/growthcharts). Instructions for using the CDC growth chart data to calculate z-scores, as well as the data files, can be found on-line.
When a child's growth is far from the norm, percentiles carry less meaning. In these situations, it is convenient to use the special charts that show the 3rd and 97th percentiles. Another approach is to quantify "less than the 5th percentile." It's hard to interpret, for example, the fact that a child's growth has gone from the 1.5 percentile to the 2.0 percentile. Then it may be best to describe growth in terms of standard deviations or z-scores. For example, a very underweight child might be described as being 2.5 standard deviations below the mean (z = -2.5). For more information about standard deviation scores, see chapter 2 in Kessler and Dawson, 1999.