Point defects, including atom vacancies, add atom and Stone-Wale defects, close to a (5, 5) single-walled carbon nanotube (SWNT) open end were studied by DFT, semiempirical PM3 methods and the empirical Brenner potential. It is found that defect stability increases at they become closer to the SWNT open end. Based on these results, a model for removing defects in a growing SWNT is proposed, where the defects diffuse to the SWNT end. Furthermore, the calculations show that the semiempirical PM3 method compares well with DFT results, and is accurate enough for studying defect formation in SWNTs. In contrast, the empirical Brenner potential yields large errors and is sometimes not even qualitatively correct.