In this paper we point out a perturbed version of a recent generalisation of the Ostrowski inequality for arbitrary divisions of a given interval [a,b]. It is shown that in certain cases the new results are better that those in [1].