Bit and bottom hole assembly (BHA) rotational speed oscillation is often encountered in North America land drilling applications. One common type of rotational speed oscillation is the well-known stick-slip, with speed variation at a frequency that is close or lower than the fundamental torsional natural frequency of the drillstring (usually less than 1Hz). Another commonly encountered rotational speed oscillation confirmed by the authors' recent studies is the high frequency torsional oscillation (HFTO), with a frequency ranging from 10's to over 100 Hz. This paper will discuss new findings on HFTO. Most of the commercially available downhole measurement tools have no difficulty picking up the low frequency stick-slip, but lack bandwidth to detect HFTO. Also it is rarely observed from the surface instruments because high frequency torsional wave can be dampened quickly when it is transferred along the drillstring. Hence, HFTO could be misinterpreted as smooth drilling until tool failure or damage occur. To understand the mechanism of HFTO, a series of downhole tests were conducted in North America land with different BHA setups and applications. An advanced downhole measurement tool was placed in the BHA to continuously record tri-axial acceleration and rotational speed at 1024 Hz sampling rate. After extensive analysis performed on the downhole high frequency data, we present in this paper many interesting findings. HFTO shifts frequency depending on different downhole conditions. There is a strong correlation between the surface parameters (RPM and SWOB) and the occurrence of HFTO. The excitation of HFTO is also related to the formation. 3D transient drilling dynamics simulation confirms the factors that increase the risk of HFTO. Based on this study, we provide recommendations to downhole measurements for capturing HFTO. With real-time HFTO detection, drilling parameters can be adjusted to reduce the chance of HFTO to avoid premature tool failures. We also provide insights for new tool development to obtain better tool life by eliminating weak spots and perform more robust measurements by selecting better sensor locations, etc. With the help of 3D transient dynamic modeling, we can gain better understanding of HFTO mechanisms through numerical simulation, making it possible to plan the BHA effectively and select drilling parameters to avoid HFTO.