Before the Magellan mission, the best estimates of the rotation rate and pole direction of Venus were derived from Earth-based radar measurements. A new determination of these rotational parameters has now been made from an analysis of Magellan radar images. Control points were selected from the north polar region and measured on full-resolution radar strips. The measurements were entered into a least squares adjustment to solve for the pole direction and rotation rate of Venus, as well as the coordinates of the control points themselves. The current data set contains 3893 measurements of 571 points on 560 radar strips. Spacecraft ephemeris errors dominate the measurement errors. One technique used to remove ephemeris errors is to adjust the averaged orbital inclination and argument of periapsis for each orbit. In a more precise technique used for selected blocks of orbits, an improved spacecraft ephemeris is computed by optimally fitting measurements of additional points at all latitudes of the radar strips together with Earth-based spacecraft radiometric tracking measurements. In a separate experiment, measurements were made of a few points common to both Venera 15/16 and Magellan images. The long time baseline between the images should have led to an accurate determination of the rotation period and pole direction of Venus. However, the measurement residuals were unexpectedly large, and these solutions are not currently considered reliable. A rotation period estimate of 243.0185¿0.0001 days was determined via the ephemeris improvement technique applied simultaneously to two overlapping orbit blocks with many common points and separated by two full Venus rotations. Using this period value, the control network computations estimated the north pole direction as right ascenson=272.76¿¿0.02¿ and declination=67.16¿¿0.01¿ in the J2000 frame. ¿ American Geophysical Union 1992