Anodic TiO 2 nanotubes (NTs) have been studied extensively for many years. However, the growth kinetics still remains unclear, because it is hardly derived by direct in situ methods. Here, an interesting approach is proposed to overcome this challenge. A combinatorial anodization was exploited to monitor the pore initiation and nanotube growth under a preformed compact surface layer (CSL). The preformed CSL and the NTs under the CSL (UCSL-NTs) were formed in fluoride-free and fluoride-containing electrolytes, respectively. The forming process of UCSL-NTs was discussed as compared with that of the general NTs, mainly focusing on the differences of current-time curves and electric charge quantity (Coulomb). The results show that pore embryos of UCSL-NTs have already been achieved under the CSL before the CSL is dissolved. There are five stages in the current-time curve of UCSL-NTs, which is significantly different from three stages of the general NTs. A new growth model, based on a comprehensive review of the existing theories, is proposed to explain the current decrease and increase. And the forming process of TiO 2 NTs is considered to be dominated by the oxide plastic flow around the oxygen bubbles.Anodic TiO 2 nanotubes (NTs) and other porous anodic oxides have attracted considerable scientific interests due to their various applications (e.g., solar energy materials, magnetic semiconductors and biosensors) 1-3 and mysterious formation mechanisms. 4,5 Different mechanisms of TiO 2 NTs have been reported in many electrochemical journals in recent years. 4-8 It is well known that field-assisted dissolution (FAD) (TiO 2 + 6F − + 4H + → [TiF 6 ] 2− + 2H 2 O) of the oxide leads to pore formation in anodic titania films, 8-10 similar to that in porous anodic alumina (PAA) films (Al 2 O 3 + 6H + → 2Al 3 + + 3H 2 O), 11-14 despite a lack of direct experimental evidence that confirms this expectation. 14 As the formation mechanism is impossible to be derived by direct in situ experimental methods, much remains to be done along these directions. 15 Garcia-Vergara et al. 16,17 proposed the field-assisted 'plastic flow' model, the constant thickness of the barrier layer is maintained by flow of oxide from the pore bottom toward the pore wall, driven by compressive stresses from electrostriction and possibly through volume expansion. 16 In fact, the plastic flow is contrary to expectations of the FAD. 16 The behavior of incorporated species in PAA is always incompatible with the FAD model. 16 The flow model has been recognized and exploited for explaining the formation of TiO 2 NTs and serrated nanochannels. 18,19 However, Zhou et al. 12 indicated that both the FAD and the flow models cannot explain the formation of gaps among nanotubes. In recent tracer studies on Ti thin films, the expansion factors increase from 1.5 to 3.0, 20,21 these findings cannot be clarified. Furthermore, anodized TiO 2 NTs have been achieved in an aqueous H 2 SO 4 solution as well as other fluoride free solutions, 12,22,23 this fact puts the flu...