Abstract Using approximation theory methods, we investigate the rate of convergence of expected L-estimates to a non-trivial associated integral limit in the d.i.d. case.