As the most prevailing two-factor authentication mechanism, smart card based password authentication has been a subject of intensive research in the past decade and hundreds of this type of schemes have been proposed. However, most of them were found severely flawed, especially prone to the smart card security breach problem, shortly after they were first put forward, no matter the security is heuristically analyzed or formally proved. In SEC’12, Wang pointed out that, the main cause of this issue is attributed to the lack of an appropriate security model to fully identify the practical threats. To address the issue, Wang presented three kinds of security models, namely Type I, II and III, and further proposed four concrete schemes, only two of which, i.e. PSCAV and PSCAb, are claimed to be secure under the harshest model, i.e. Type III security model. However, in this paper, we demonstrate that PSCAV still cannot achieve the claimed security goals and is vulnerable to an offline password guessing attack and other attacks in the Type III security mode, while PSCAb has several practical pitfalls. As our main contribution, a robust scheme is presented to cope with the aforementioned defects and it is proven to be secure in the random oracle model. Moreover, the analysis demonstrates that our scheme meets all the proposed criteria and eliminates several hard security threats that are difficult to be tackled at the same time in previous scholarship, which highly indicates the settlement of an open problem raised by Madhusudhan and Mittal in 2012. Beyond our cryptanalysis of current schemes and our proposal of the new scheme, the proposed adversary model and criteria set provide a benchmark for the systematic evaluation of future two-factor authentication proposals.