An upper bound is established for the magnitude of the truncation error incurred when a real-valued finite energy signal that is bandlimited to

, is approximated by

terms from its Shannon sampling series expansion with the samples taken at the integer points. The derived bound is an improvement on known results.