Objective: To investigate the application of high-frequency ultrasound in dermabrasion of patients with deep partial-thickness burns. Methods: Twenty-six patients with deep partial-thickness burns conforming to the study criteria were hospitalized in our unit from March 2015 to March 2016. Patients were all performed with dermabrasion. The structure of skin tissue and blood flow signals of uninjured side and wounds before dermabrasion, and those of wounds immediately post dermabrasion and on post dermabrasion day (PDD) 1, 3, 5, 7, 10, 14, and 21 were detected with high-frequency ultrasound, and the percentage of blood flow signals was calculated. According to the results of comparison between percentage of blood flow signals of wounds and that of normal skin before dermabrasion, patients were divided into no significant decrease group (NSD, n=19) and significant decrease group (SD, n=7). Wound healing time of patients in two groups was recorded. Data were processed with analysis of variance of repeated measurement, LSD test, t test and Chi-square test. The correlation between the percentage of blood flow signals of wounds before dermabrasion and wound healing time of 26 patients were analyzed by Spearman correlation analysis. Results: (1) Epidermis of normal skin of patients in two groups before dermabrasion showed continuous smooth linear hyperecho, which was stronger than that of dermis, and boundary of dermis and subcutaneous tissue showed stronger discontinuous linear echo than that of dermis, which gradually transited to subcutaneous tissue. In group NSD, epidermis of wound of patients before dermabrasion showed intermittent rough linear echo, which was weaker than that of normal skin epidermis, and there was no obvious abnormity of boundary between dermis and subcutaneous tissue. Immediately post dermabrasion and on PDD 1, no linear hyperecho of epidermis was observed, showing complete attrition of epidermis, and the echo of dermis and subcutaneous tissue had no obvious change as compared with that before dermabrasion, with flat surface of dermis and partly abraded superficial-dermis but relatively well preserved dermal tissue in whole. The epidermis showed discontinuous linear hyperecho, and epidermis was discontinuously regenerated on PDD 3 and 5. Partial continuous linear hyperecho was detected in the epidermis, showing partial continuous regeneration of epidermis on PDD 7 and 10. The regenerated epidermis was thicker than normal skin epidermis and showed rough linear hyperecho with non-uniform thickness on PDD 14. The regenerated epidermis was thicker than normal skin epidermis and showed rather smooth linear hyperecho with uniform thickness on PDD 21. In group SD, the structure of epidermis and dermis of wound of patients before dermabrasion, immediately post dermabrasion, and on PDD 1 was similar to that in group NSD, but the echo of boundary of dermis and subcutaneous tissue was weakened in different degrees. There was no linear hyperecho of epidermis, showing no epidermis was regenerated on PDD 3 and 5. Intermittent regeneration of epidermis appeared on PDD 7 and 10 with intermittent linear hyperecho. Partial continuous linear hyperecho was detected in the epidermis, showing partial continuous regeneration of epidermis on PDD 14 and 21. (2) The percentages of blood flow signals of wounds of patients in group NSD before dermabrasion, immediately post dermabrasion, and on PDD 1 were (3.1±1.3)%, (6.5±2.0)%, and (5.3±1.9)% respectively, higher than those in group SD [(0.9±1.1)%, (3.5±1.3)%, and (3.6±0.9)% respectively, P<0.05 or P<0.01]. The percentages of blood flow signals of wounds of patients in two groups were similar at the other time points (with P values above 0.05). Compared with the percentage of normal skin in the same group [(3.2±0.7)%], the percentages of blood flow signals of wounds of patients in group NSD were significantly increased immediately post dermabrasion and on PDD 1 (with P values below 0.01) but had no significant change at the other time points (with P values above 0.05). The percentage of blood flow signals of wounds of patients before dermabrasion in group SD was significantly lower than that of normal skin in the same group [(2.8±0.6)%, P<0.01]. The percentage of blood flow signals of wounds of patients in group SD was close to that of normal skin in the same group at each time point post dermabrasion (with P values above 0.05). (3) The wound healing time of patients in group NSD was (16.2±2.5) d, lower than that in group SD [(30.9±2.9) d, t=12.67, P<0.01]. There was obvious negative correlation between the percentage of blood flow signals of wounds before dermabrasion and wound healing time of 26 patients (r=-0.77, P<0.01). Conclusions High-frequency ultrasound is a good way to observe the imaging features of wounds in patients with deep partial-thickness burns before and after dermabrasion, and it can provide objective imaging evidence for the performance of dermabrasion in patients with deep partial-thickness burns.