Ionosondes underwent significant improvements since their invention in 1924 (Breit and Tuve, 1926). At the beginning, ionospheric sounders measured only vertical profile of electron concentration. Contemporary modern digital ionosondes usually provide ionospheric drift measurement as a part of the routine monitoring in addition to the classical vertical ionospheric sounding (ionogram measurement). Ionospheric plasma motion monitoring has a large potential to improve our understanding of an ionospheric dynamics. However the measurement quality is highly variable. In this paper we deal with conditions and assumptions necessary for the correct drift velocity estimation, particularly differences in measurements for E and F regions. Correct and accurate drift velocity estimation requires recording of a sufficient number of reflection points during the measurement. We discuss how to obtain good quality drift data and how to estimate data quality. Large number of recorded drift measurements are eliminated from further use for drift velocity estimation due to low number of detected reflection points. Our analysis done for a large dataset strongly supports the idea that even low number of recorded reflection points may indicate an expressive horizontal stratification of the ionosphere. For our study we used drift data collected in 2006 from a mid-latitude station Pruhonice. Data were collected during a period of low geomagnetic and solar activity. We show statistical properties of drift velocity components for both E and F regions. We illustrate the influence of geomagnetic activity on drift velocities on maximal daily value of F region horizontal component (geomagnetic activity is represented by Kp). We also show statistics for poor-quality data associated with horizontal stratification of ionosphere.