We examine the closure of the current plate motion circuit between the African, North American, and Eurasian plates to test whether these plates are rigid and whether the Gloria fault is an active transform fault. We also investigate the possible existence of microplates that have been previously proposed to lie along these plate boundaries, and compare the predicted direction of motion along the African‐Eurasian plate boundary in the Mediterranean with the direction of slip observed in earthquakes. From marine geophysical data we obtain 13 transform fault azimuths and 40 3‐m.y.‐average spreading rates, 34 of which are determined from comparison of synthetic magnetic anomaly profiles to ∼140 observed profiles. Slip vectors from 32 earthquake focal mechanisms further describe plate motion. Detailed magnetic surveys north of Iceland provide 11 rates in a region where prior plate motion models had few data. Magnetic profiles north of the Azores triple junction record a rate of 24 mm/yr, 4 mm/yr slower than used by prior models. Gloria and Sea Beam surveys accurately measure the azimuths of seven transform faults; our plate motion model fits six of the seven within 2°. Two transform faults surveyed by Gloria side scan sonar lie near FAMOUS area transform faults A and B and give azimuths 13° clockwise of them. Because recent studies show that short‐offset transforms, such as transforms A and B, are in many places oblique to the direction of plate motion, we exclude azimuths from transforms with less than 35‐km offset. The best fitting and closure‐enforced vectors fit the data well, except for a small systematic misfit to the slip vectors: On right‐lateral slipping transforms, slip vectors tend to be a few degrees clockwise of plate motion and mapped fault azimuths, whereas on left‐lateral slipping transforms, slip vectors tend to be a few degrees counterclockwise of plate motion and mapped fault azimuths. We search the long Eurasia‐North America boundary for evidence of an additional plate, but find no systematic misfits to the data. In particular, if a Spitsbergen plate exists and moves relative to Eurasia, its motion is less than 3 mm/yr. An Africa‐Eurasia Euler vector determined by adding the Eurasia‐North America and Africa‐North America Euler vectors is consistent with the Gloria fault trend and with slip vectors from eastern Azores‐Gibraltar Ridge focal mechanisms. A small circle, centered at the Africa‐Eurasia closure‐enforced pole, fits the trace of the Gloria fault. The model in which closure was enforced predicts ∼4 mm/yr slip across the Azores‐Gibraltar Ridge, and west‐northwest convergence near Gibraltar, ∼45° more oblique than suggested by a recent model based on compressive axes of focal mechanisms. Moreover, our model predicts directions of plate motion that agree well with northwest trending slip vectors from thrust earthquakes between Gibraltar and Sicily. Because closure‐enforced vectors fit the data nearly as well as the best fitting vectors, we conclude that the data are consistent with a rigid plate model and with the Gloria fault being a transform fault.