A decoupled slow speed friction torque model has been developed to predict the running torque of a double-arched angular contact bearing when operating as a four-point, as well as a two-point contact bearing. The load distribution model from Amasorrain et al. (2003) and the kinematics model developed by Leblanc and Nelias (2007) have been combined after ignoring centrifugal and gyroscopic effects, a valid assumption for slow speed operation. Results from the model are compared with previous literature data, as well as with tests done on a specially-developed friction torque rig. The comparison with the literature results was done for two specific cases: (i) when only one out of the two contact lines in the four-point contact bearing is active (effectively creating a two-point contact bearing); and (ii) where both contact lines in the four-point contact bearing are active. Further, the comparison was done with a custom-built friction torque rig with FAG QJ309 bearings, again for two cases: (i) bearings mounted with a specific clearance (two-point contact); and (ii) bearings mounted with larger size balls to obtain interference (four-point contact). All tests were performed at low speeds. The sliding friction, which is an important input to the friction torque model, is carefully measured on ball-on-plate test using the same interface roughness, speed and contact pressure conditions as seen in the QJ309 friction test. The model comparison with experimental results is covered. The comparison is found to be encouraging, with the RMS difference being less than 7% between the model and experimental data for a four point contact.