You should be able to use gname. Depending on how your data is organized, you might need to repmat the labels. For example
load cities
boxplot(ratings)
gname(repmat(names,9,1))
In the cities example data set, the names correspond to the rows of the matrix ratings. The boxplot uses each column of ratings as the categories. So if you click on a data point in the 7th column, gname is trying to (linearly) index into element 2000 (or whatever). So doing repmat on the names simply repeats the labels for each of the 9 columns.
