Objective: To compare the short-term and long-term outcomes between transanal total mesorectal excision (taTME) and laparoscopic total mesorectal excision (laTME) for mid-to-low rectal cancer and to evaluate the learning curve of taTME. Methods: This study was a retrospective cohort study. Firstly, consecutive patients undergoing total mesorectal excision who were registered in the prospective established database of Division of Colorectal Diseases, Department of General Surgery, Peking Union Medical College Hospital during July 2014 to June 2020 were recruited. The enrolled patients were divided into taTME and laTME group. The demographic data, clinical characteristics, neoadjuvant treatment, intraoperative and postoperative complications, pathological results and follow-up data were extracted from the database. The primary endpoint was the incidence of anastomotic leakage and the secondary endpoints included the 3-year disease-free survival (DFS) and the 3-year local recurrence rate. Independent t-test for comparison between groups of normally distributed measures; skewed measures were expressed as M (range). Categorical variables were expressed as examples (%) and the χ(2) or Fisher exact probability was used for comparison between groups. When comparing the incidence of anastomotic leakage, 5 variables including sex, BMI, clinical stage evaluated by MRI, distance from tumor to anal margin evaluated by MRI, and whether receiving neoadjuvant treatment were balanced by propensity score matching (PSM) to adjust confounders. Kaplan-Meier curve and Log-rank test were used to compare the DFS of two groups. Cox proportional hazard model was used to analyze and determine the independent risk factors affecting the DFS of patients with mid-low rectal cancer. Secondly, the data of consecutive patients undergoing taTME performed by the same surgical team (the trananal procedures were performed by the same main surgeon) from February 2017 to March 2021 were separately extracted and analyzed. The multidimensional cumulative sum (CUSUM) control chart was used to draw the learning curve of taTME. The outcomes of 'mature' taTME cases through learning curve were compared with laTME cases and the independent risk factors of DFS of 'mature' cases were also analyzed. Results: Two hundred and forty-three patients were eventually enrolled, including 182 undergoing laTME and 61 undergoing taTME. After PSM, both fifty-two patients were in laTME group and taTME group respectively, and patients of these two groups had comparable characteristics in sex, age, BMI, clinical tumor stage, distance from tumor to anal margin by MRI, mesorectal fasciae (MRF) and extramural vascular invasion (EMVI) by MRI and proportion of receiving neoadjuvant treatment. After PSM, as compared to laTME group, taTME group showed significantly longer operation time [(198.4±58.3) min vs. (147.9±47.3) min, t=-4.321, P<0.001], higher ratio of blood loss >100 ml during surgery [17.3% (9/52) vs. 0, P=0.003], higher incidence of anastomotic leakage [26.9% (14/52) vs. 3.8% (2/52), χ(2)=10.636, P=0.001] and higher morbidity of overall postoperative complications [55.8%(29/52) vs. 19.2% (10/52), χ(2)=14.810, P<0.001]. Total harvested lymph nodes and circumferential resection margin involvement were comparable between two groups (both P>0.05). The median follow-up for the whole group was 24 (1 to 72) months, with 4 cases lost, giving a follow-up rate of 98.4% (239/243). The laTME group had significantly better 3-year DFS than taTME group (83.9% vs. 73.0%, P=0.019), while the 3-year local recurrence rate was similar in two groups (1.7% vs. 3.6%, P=0.420). Multivariate analysis showed that and taTME surgery (HR=3.202, 95%CI: 1.592-6.441, P=0.001) the postoperative pathological staging of UICC stage II (HR=13.862, 95%CI:1.810-106.150, P=0.011), stage III (HR=8.705, 95%CI: 1.104-68.670, P=0.040) were independent risk factors for 3-year DFS. Analysis of taTME learning curve revealed that surgeons would cross over the learning stage after performing 28 cases. To compare the two groups excluding the cases within the learning stage, there was no significant difference between two groups after PSM no matter in the incidence of anastomotic leakage [taTME: 6.7%(1/15); laTME: 5.3% (2/38), P=1.000] or overall complications [taTME: 33.3%(5/15), laTME: 26.3%(10/38), P=0.737]. The taTME was still an independent risk factor of 3-year DFS only analyzing patients crossing over the learning stage (HR=5.351, 95%CI:1.666-17.192, P=0.005), and whether crossing over the learning stage was not the independent risk factor of 3-year DFS for mid-low rectal cancer patients undergoing taTME (HR=0.954, 95%CI:0.227-4.017, P=0.949). Conclusions: Compared with conventional laTME, taTME may increase the risk of anastomotic leakage and compromise the oncological outcomes. Performing taTME within the learning stage may significantly increase the risk of postoperative anastomotic leakage.