The identification of discrete-time (DT) models, under assumption that the main errors come from in accurateness of measurements, and the information processing is relatively precise, is researched. It becomes, that for small sampling periods the estimation accuracy of the shift operator (SO) and delta operator (DO) models is then comparable. Two methods of the increase of identification accuracy are proposed; the first based on the increase of the number of measured signals and realised by additional intermediate measurement (s); the second based on the decrease of the number of estimated parameters and realised by using the information about known parameters. The first method results from the original observation that the separated by an additional intermediate measurement parts of continuous-time plant may be described by a DT transfer functions.