The word association task is one of the most important techniques to explore knowledge structures in human beings. It has been applied to the educational, clinical, and business research areas. Thus, establishing a comprehensive association norm that contains enough staple words as stimuli is a necessity. However, only a few association norms were established in Chinese, and these norms were often only prepared for words of particular categories. Hence, the goal of the present study was to establish a larger Chinese association norm covering more stimulated words. In the preliminary experiment, 30 subjects were asked to rate the level of abstraction of 900 frequent and 900 infrequent Chinese words with a 7-point scale. After deleting the words with large variability in abstraction ratings, there were 1800 Chinese words remaining. From this pool, 200 words from each combination of high/medium/low abstraction levels and frequent/infrequent words (six combinations in total) were chosen, resulting in a total of 1200 words chosen as word stimuli in the formal study. We also asked 104 subjects to rate the valence degree on these word stimuli in order to let follow-up researchers select the stimulus from the norm reported by the present study. In the formal study, 1200 subjects were recruited, and each performed a free association task with only 200 words among the word stimuli. An average of 200 responses were collected for each cue. Commonality, idiosyncratic index, and associated reference index of each stimulus word were analyzed. The result showed that commonality of low frequency or high abstraction words was higher than the commonality of high frequency or low abstraction words. It further indicated that low frequency or high abstraction words yielded more consistent and convergent associative responses. The association norms from our study could help future researchers explore the Chinese semantic network and manipulate experimental stimuli more carefully.