The balance of global angular momentum is examined in a long time series of NCEP/NCAR Reanalysis data. Based on the 1968-96 period, a systematic bias of about -10 Hadleys is obtained in the total torque in all seasons. The sum of the frictional and mountain torques contributes about a half of the bias in the northern summer and autumn. The torque due to parameterized gravity wave drag contributes the other half of the negative bias in summer, but the whole of it in winter. In the absence of the gravity wave torque, the seasonal and annual budget in the NCEP/NCAR Reanalysis from 1968-96 broadly agree with the classical estimates by Newton (1972) using early observations. However, Newton's results showed a more pronounced seasonal cycle for the mountain torque (with a more positive seasonal mean in autumn, and more negative in spring), and a less negative frictional torque in summer. It remains to be determined whether these differences imply a long-term change of the balance of global angular momentum, or they reflect some unique characteristics of the NCEP GCM used for reanalysis.