This paper describes the submission from the University of Helsinki to the\nshared task on cross-lingual dependency parsing at VarDial 2017. We present\nwork on annotation projection and treebank translation that gave good results\nfor all three target languages in the test set. In particular, Slovak seems to\nwork well with information coming from the Czech treebank, which is in line\nwith related work. The attachment scores for cross-lingual models even surpass\nthe fully supervised models trained on the target language treebank. Croatian\nis the most difficult language in the test set and the improvements over the\nbaseline are rather modest. Norwegian works best with information coming from\nSwedish whereas Danish contributes surprisingly little.\n