Why does vmstat terminate with signal 8, Arithmetic exception on Red Hat Enterprise Linux 5?

Issue

When running vmstat 1 for about 20 minutes it terminates with a signal 8 Arithmetic exception.

When a core dump is captured and analysed, a backtrace similar to the following is reported:

Thread 1 (process 15693):
#0  0x00000000004021f5 in main (argc=<value optimized out>, argv=<value optimized out>) at vmstat.c:255