A number of different Euler poles have been proposed by various authors during the past 20 years to describe the relative motion of Caribbean (CA) and North American (NA) plates. These different solutions are a consequence of the small number of focal mechanism solutions of earthquakes, the short length of the plate boundaries of the Caribbean that consists of an active spreading center, the complicated and multibranched fault systems of several parts of the plate boundaries, and biases introduced in using slip vectors along the Middle American Trench to close plate motion circuits in computing CA‐NA Euler poles. Slip vectors from focal mechanisms are now the strongest constraints on the direction of current plate motion. We use 66 slip vectors derived from 117 focal mechanisms of shallow earthquakes located on or near the boundaries between the Caribbean and either the NA or South American (SA) plates to calculate a best fitting, present‐day rotation pole for CA‐NA. It is located at 68.4°N, 126.3°W, near Barrow, Alaska. In making this calculation, we effectively extended the length of the plate boundary considered in the calculation by using data from the CA‐SA plate boundary and rotating them to the CA‐NA reference frame using the Nuvel 1 pole for NA‐SA motion and 20 mm/yr as the rate of CA‐SA plate motion along the southeastern Caribbean. The result is not very sensitive to CA‐SA rates. The azimuth of plate motion is about N70°E along most of the CA‐NA boundary. For example, along the Puerto Rico Trench, several focal mechanisms indicate highly oblique motion with a small component of plate convergence along shallow, south dipping thrust faults. The highly oblique plate motion along the inner wall of the Puerto Rico Trench is indicative of strong coupling of that forearc with the Caribbean plate and of the potential of that zone to generate great earthquakes. Slip vectors are very consistent with one another in areas in which at least one of the interacting plates is oceanic. Focal mechanisms and the distribution of earthquakes exhibit greater scatter, however, in areas such as Hispaniola where continental or thick island arc lithosphere is being deformed and is present on both sides of that complex plate boundary. In addition, other mechanisms indicate northerly subduction of the Caribbean beneath southeastern Hispaniola along the western Muertos Trough, normal faulting to the east of the Lesser Antilles associated with the bending of the downgoing plate, and normal faulting or reverse faulting in more localized areas near the plate boundary. As part of the analysis, we also calculated Euler poles for CA‐SA motion at 52°N, 81°W and Cocos‐Caribbean (CO‐CA) at 22°N, 120°W. These poles are consistent with recent models of transtensional development since about 10 m.y. ago along the plate boundary zone of the southeastern Caribbean and with CO‐CA motion that involves subduction along the Middle America Trench as well as right‐lateral, strike‐slip faulting and extension along the volcanic zone of Central America. The Euler pole for CO‐CA is also consistent with recent