INTRODUCTION: The ability of fingers to rapidly rewarm following cold exposure is a possible indicator of cold injury protection. We categorized the post-cooling hand-rewarming responses of men before and after participation in 15 mo of military training in a cold environment in northern Sweden to determine: 1) if the initial rewarming category was related to the occurrence of local cold injury during training; and 2) if cold training affected subsequent hand-rewarming responses. METHODS: Immersion of the dominant hand in 10 degrees C water for 10 min was performed pre-training on 77 men. Of those, 45 were available for successful post-training retests. Infrared thermography monitored the dorsal hand during 30 min of recovery. Rewarming was categorized as normal, moderate, or slow based on mean fingertip temperature at the end of 30 min of recovery (TFinger,30) and the percentage of time that fingertips were vasodilated (%VD). RESULTS: Cold injury occurrence during training was disproportionately higher in the slow rewarmers (four of the five injuries). Post-training, baseline fingertip temperatures and cold recovery variables increased significantly in moderate and slow rewarmers: TFinger30 increased from 21.9 +/- 4 to 30.4 +/- 6 degrees C (Moderate), and from 17.4 +/- 0 to 22.3 +/- 7 degrees C (Slow); %VD increased from 27.5 +/- 16 to 65.9 +/- 34% (Moderate), and from 0.7 +/- 2 to 31.7 +/- 44% (Slow). CONCLUSIONS: Results of the cold recovery test were related to the occurrence of local cold injury during long-term cold-weather training. Cold training itself improved baseline and cold recovery in moderate and slow rewarmers.