This study examined the psychometric properties of the Chinese version of the Fagerstrom Test for Nicotine Dependence (FTND). Criterion validity, factorial structure, and construct reliability were conducted using PRELIS 2 and LISREL 8.7 software. Two hundred and forty-five adult smokers were recruited from April to August 2005 in Taiwan. Because some items of the FTND are dichotomous and ordinal, biserial and polyserial correlations between the items and saliva cotinine levels were utilized. Confirmatory factor analyses (CFA) were performed to compare the relative fit of three competing models. Computation of the construct reliability was necessitated by a lack of tau-equivalence. The results demonstrated that the Chinese version has satisfactory validity and reliability. One item, "Smoke more in the morning," had low correlations with saliva cotinine. CFA found the items of the Chinese version were best modeled as two correlated factors without a cross-loading. These findings differ from the previous research testing the English version. Studies addressing the specific features of cigarette dependence in different socio-cultural contexts are needed.