The availability of Web 2.0 technologies has made it easy for information users to express their own opinions and access other people's opinions on the Web. We are interested in understanding how opinions expressed in one way by one group compare to opinions expressed in another way by another group, especially in a different language. We have done reasonably well at finding opinionated English mailing lists and blogs, so we started to work on Chinese opinion classification. This paper reports on experiments with a recently released opinion classification test collection for Chinese sentences. Term-scale evidence from a large lexicon and from character-based estimation of semantic orientation for unknown words was used to construct classifiers for subjectivity and polarity that are somewhat more accurate than the best previously reported results. Subjectivity density and the relative predominance of terms with positive and negative semantic orientation were found to be useful features, and appropriate handling of negation was found to be important. With bilingual opinion classification techniques, we can help users find and compare opinions about a topic in two languages.