Oceanic transform faults are often thought to 'guide' plate motions. As such, transform azimuth data have been an important component in global kinematic models such as NUVEL-1A. With the advent of space-geodesy most global plate motion models are now solely determined using geodetic velocity measurements and are thus independent of oceanic ridge and transform data. For some plates significant differences between geodetic-derived motions and NUVEL-1A predictions are found. However, a systematic global analysis of possible discrepancies between observed and expected transform azimuths, or a joint inversion of transform faults and geodetic velocities, does not exist to date. Elsewhere, we have presented a global kinematic model that describes plate motions as well as the velocity gradient tensor field within plate boundary zones. The velocity gradient tensor field is estimated through a least-squares fit between model and observed velocity vectors, and model and observed strain rates. Observed velocity vectors consist of 3000 geodetic velocities measured on stable plates as well as within plate boundary zones. Observed strain rates are derived from Quaternary fault slip rate data in Asia Here, we first present a comparison between azimuth predictions from this global model and observed azimuth directions. This analysis should highlight problematic areas, such are known to exist, for instance, west of the Macquarie triple junction. Second, we present an alternative model to the global model by incorporating ocean transforms as an additional no-length-change direction constraint in the inversion of geodetic velocities. We evaluate the model change due to this additional constraint in terms of the misfit between model and observed velocities located on the plates as well as between the predicted and modelled transform azimuths. Of special attention is the impact of the geographic distribution of both geodetic velocities on the plate and transform faults around the plate. The questions that we hope to address are: Are present-day plate motions and transform azimuths mutually consistent ? Does the inclusion of a no-length-change constraint along ocean transforms improve our global kinematic model ? Are there local discrepancies for individual plates, and how can these be explained ?