This paper analyses some aspects of the eye movement behaviour of readers of Thai and Chinese.The main focus is on readers landing site distributions on words and how these are affected by the lack of clear word boundary information due to the absence of inter-word spaces.Empirical evidence from Thai and Chinese readers suggest that readers can relatively accurately target word cen- tres.We make the case that this accuracy can be accounted for by a default targetting model (effectively, the prior landing site dis- tribution, in Bayesian terms) modulated by statistical cues about word beginnings available from word-initial character frequencies.