The objective of this study is to evaluate clinical and radiological outcomes of unstable or intra-articular Colles' fractures using the Pennig model of external fixation. The Pennig model of dynamic external fixation was used on 88 patients with 92 Colles' fractures. In 35 wrists, exercise was not performed during external fixation. In 57 wrists, exercise was initiated three weeks after surgery. Regarding radiological measurements, the degrees of volar tilt (VT), radial inclination (RI), radial length (RL) and ulnar variance (UV) were serially measured and evaluated. The clinical results were evaluated according to Sarminento's criteria. Favourable therapeutic outcomes (excellent and good) were obtained in 95.6% of the cases. However, when complicated by die-punch fractures, the outcome tended to be poorer. Postoperative re-displacement was often observed in patients over 50 years old. UV tended to increase and RL tended to decrease in the early motion group after the initiation of exercise. Neither VT nor RI significantly changed in either group during the postoperative observation period. Of patients in whom bone union of ulnar styloid fractures was not achieved, 32.2% complained of ulnar pain. Moreover, osteoarthritic changes were observed in two of 11 wrists with a step off of 1 to 2 mm or larger. In order to maintain the reduction position, it was important to determine the appropriate time for initiating exercise depending on the degree of fixation in each individual patients. The step off of the articular surface should be reduced to less than 1 mm and reductive fixation has to be performed additionally when complicated by ulnar styloid fractures.