The management of radial nerve palsy associated with fractures of the shaft of the humerus has been disputed for several decades. This study has systematically reviewed the published evidence and developed an algorithm to guide management. We searched web-based databases for studies published in the past 40 years and identified further pages through manual searches of the bibliography in papers identified electronically. Of 391 papers identified initially, encompassing a total of 1045 patients with radial nerve palsy, 35 papers met all our criteria for eligibility. Meticulous extraction of the data was carried out according to a preset protocol. The overall prevalence of radial nerve palsy after fracture of the shaft of the humerus in 21 papers was 11.8% (532 palsies in 4517 fractures). Fractures of the middle and middle-distal parts of the shaft had a significantly higher association with radial nerve palsy than those in other parts. Transverse and spiral fractures were more likely to be associated with radial nerve palsy than oblique and comminuted patterns of fracture (p < 0.001). The overall rate of recovery was 88.1% (921 of 1045), with spontaneous recovery reaching 70.7% (411 of 581) in patients treated conservatively. There was no significant difference in the final results when comparing groups which were initially managed expectantly with those explored early, suggesting that the initial expectant treatment did not affect the extent of nerve recovery adversely and would avoid many unnecessary operations. A treatment algorithm for the management of radial nerve palsy associated with fracture of the shaft of the humerus is recommended by the authors.