Cryptanalytic time-memory trade-offs (TMTO) are well-knowntools available in any security expert toolbox. They have been used tobreak ciphers such as A5/1, but their efficiency to crack passwords madethem even more popular in the security community. While symmetrickeys are generated randomly according to a uniform distribution, pass-words chosen by users are in practice far from being random, as con-firmed by recent leakage of databases. Unfortunately, the technique usedto build TMTOs is not appropriate to deal with non-uniform distribu-tions. In this paper, we introduce an efficient construction that consists inpartitioning the search set into subsets of close densities, and a strategyto explore the TMTOs associated to the subsets based on an interleavedtraversal. This approach results in a significant improvement comparedto currently used TMTOs. We experimented our approach on a classicalproblem, namely cracking 7-character NTLM Hash passwords using analphabet with 34 special characters, which resulted in a 16 × speedupover rainbow tables, which are considered as the most efficient variant oftime-memory trade-offs.