1: \begin{abstract}

2: For non-uniformly hyperbolic dynamical systems we consider the time 

3: series of maxima along typical orbits. Using ideas based upon 

4: quantitative recurrence time statistics we prove convergence

5: of the maxima (under suitable normalization) to an extreme value distribution, 

6: and obtain estimates on the rate of convergence. We show that our results are applicable to a range

7: of examples, and include new results for Lorenz maps, certain partially hyperbolic 

8: systems, and non-uniformly expanding systems with sub-exponential decay of correlations. 

9: For applications where analytic results are not readily available

10: we show how to estimate the rate of convergence to an extreme 

11: value distribution based upon numerical information of the quantitative recurrence statistics.

12: We envisage that such information will lead 

13: to more efficient statistical parameter estimation schemes based upon the block-maxima method.

14: 

15: \end{abstract}

16: