In this paper we describe our approach at the PAN 2010 plagiarism detection competition. We refer to the system we have used in PAN'09. We then present the improvements we have tried since the PAN'09 competition, and their impact on the results on the development corpus. We describe our experiments with intrinsic plagiarism detection and evaluate them. We then discuss the computational cost of each step of our implementation, including the performance data from two different computers.