Anastomotic leakage (AL) is the most important and one of the most serious complications after colorectal resections with primary anastomosis. Any factors that contribute to increase the risk of AL should be identified and--if possible--eliminated. Non-steroidal anti-inflammatory drugs (NSAIDs) are often used for treating pain after surgical procedures, among these also colorectal resections. The objective of this Ph.d. thesis was to investigate whether the use of NSAIDs in the postoperative period increases the risk of AL, and investigate the effect on pathophysiological mechanisms. In order to achieve this, the following studies were performed. Study I was a retrospective, case-control study in 75 patients undergoing laparoscopic colorectal resection for colorectal cancer. 33 of these patients received the NSAID diclofenac in the postoperative period; the remaining 42 did not receive any NSAID. There were significantly more ALs among the patients receiving diclofenac (7/33 vs. 1/42, p=0.018). In uni- and multivariate logistic regression analyses, diclofenac was the only factor associated with increased AL rate. This study functioned as a hypothesis generating study and laid the ground for the subsequent studies. Study II was an experimental, randomized, case-control study in 32 Wistar rats. The rats had a colonic anastomosis performed and were randomized to diclofenac or placebo treatment. After three days, the rats were sacrificed and the anastomoses were harvested. First, the anastomotic strengths were tested by longitudinal; subsequently, the levels of the enzyme cyclooxygenase-2 (COX-2) in the anastomotic tissues were measured. There was no difference among the groups with regard to anastomotic strength, but the animals treated with diclofenac had significantly lower COX-2 levels (median (range) 1.30 (0.42-3.31) ng/mg vs. 2.44 (0.88 - 18.94) ng/mg, p<0.001). This study showed that the used dose of diclofenac was sufficient and relevant, but did not show a direct damaging effect on the anastomoses due to NSAID treatment. Study III was also an experimental, randomized, case-control study. This time round, 60 Wistar rats were included. Again, colonic anastomoses were performed and the rats were randomized to diclofenac or placebo. Also, expanded polytetrafluoruethylene (ePTFE) tubes were placed under the skin of the rats. In this material, substituents of connective tissue accumulate and the amount of accumulation can be measured. After 7 days, the rats were sacrificed and, again, anastomotic strengths were measured along with collagen content in the ePTFE tubes. Anastomotic strength was similar in the two groups while collagen accumulation was significantly decreased among the rats treated with diclofenac (median (i.q.r.) 0.29 (0.13-0.47) vs. 0.47 (0.28-0.62) mcg/mg, p = 0.03). This study for the first time showed that NSAID inhibit subcutaneous collagen formation and that this formation is reversely correlated to anastomotic strength. This information can be used in further studies in this subject. Study IV was the final experimental case-control study in 40 Wistar rats. This time, in order to more easily extrapolate experimental results to daily clinical life, the colonic anastomoses were sutured with the same type of suture material as used in the clinical setting. Thus, half the anastomoses was performed with resorbable suture; the other half with non-resorbable suture. None of the rats received NSAID. The breaking strength was compared and found similar in the two groups. This study showed that experimental studies can be optimized in order to make comparisons and extrapolations to the clinical setting easier. Study V was a database study based on data from the Danish Colorectal Cancer Group's (DCCG) prospective database and electronically registered medical records. From the database information on demographic, surgical and postoperative variables (including AL) were provided. Information on NSAID consumption was retrieved by individual searches in the patients' medical records. Based on these data, uni- and multivariate logistic regression analyses were performed. These analyses identified NSAID treatment in the postoperative period as an individual risk factor for AL. Other risk factors identified were consistent with the available literature. The detrimental effect of the NSAIDs are possibly due to an effect on collagen metabolism leading to weakened tissue around the anastomosis and/or on the risk of thrombosis formation leading to more thromboses in the vessels supplying the anastomosis, thereby limiting anastomotic blood flow. In conclusion, the studies included in this thesis have elucidated some of the physiological and pathophysiological mechanisms involved in anastomotic healing and leakage, and furthermore have shown that the use of NSAIDs in the postoperative period increase the risk of AL in patients undergoing colorectal surgery with primary anastomosis. Based on the findings in these studies, and based on existing knowledge, it is recommended that NSAIDs be abandoned after colorectal resection with primary anastomosis. It should be investigated whether the NSAIDs are also harmful to other types of anastomoses and after other surgical procedures where early tissue healing is crucial.