In the present paper the author gives a component-wise error estimate for an approximate solution obtained by practicing an iterative method on a floating-point system. Further, in case of Newton's method, he gives a more precise error estimate and also a simple convenient rule for stopping the iterative process.