If what you are looking for is a metric, that is a formal method of measuring parse-tree similarity which delivers numbers that correlate well with our intuitive perceptions of similarity in this area, then I would recommend a paper by me and Anna Babarczy, "A test of the leaf-ancestor metric for parse accuracy", in _J. of Natural Language Engineering_ vol. 9, pp. 365-80, 2003: this gives experimental evidence to support the claim that our "leaf-ancestor metric" is superior to what at that time was the almost universally used "Parseval" metric (in its various versions). However, if you are looking for a piece of software that performs the calculations (for some particular metric), then I don't think I can help; we wrote our own.